Smart needle systems and related methods

Abstract

Smart needle systems and related methods, such as smart injection needle system for accurate characterization of cell types, concentrations, and viability, are generally provided. In some embodiments, fluidic components (e.g., configured to interface with a medical device such as a syringe or in-line flow system) are provided.

Description

[0001] SMART NEEDLE SYSTEMS AND RELATED METHODS

[0002] RELATED APPLICATIONS

[0003] This application claims priority under 35 U.S.C. § 119(e) to U.S. Provisional Patent Application Serial No. 63 / 726,171, filed November 27, 2024, and entitled “SMART NEEDLE SYSTEMS AND RELATED METHODS,” which is incorporated herein by reference in its entirety for all purposes.

[0004] TECHNICAL FIELD

[0005] Smart needle systems and related methods, such as smart injection needle system for accurate characterization of cell types, concentrations, and viability, are generally described.

[0006] BACKGROUND

[0007] Cell therapies have transformed the possibilities for patient care across a range of conditions including neurologic, oncologic, endocrinologic, cardiac and autoimmune conditions. Central to the efficacy of intervention is the viability and dose or cell count at the time of cell transplantation. However, current approaches are challenged by involving adjacent in parallel analyses by employing bulky assays and instrumentation. Accordingly, new devices and methods are needed.

[0008] SUMMARY

[0009] The subject matter of the present disclosure involves, in some cases, interrelated products, alternative solutions to a particular problem, and / or a plurality of different uses of one or more systems and / or articles.

[0010] In one aspect, in-line flow systems are provided. In some embodiments, the system comprises a fluidic channel configured to receive a fluid comprising a plurality of cells, a sensor associated with the fluidic channel, the sensor comprising an impedance detector comprising two or more electrodes in electrical communication with the fluidic channel and an optical component proximate the fluidic channel, wherein the optical component comprises one or more photodetectors, wherein the sensor is configured to determine one or more characteristics of the plurality of cells based upon a signal from the impendence detector and / or the one or more photodetectors as the fluid flows through the fluidic channel.

[0011] In another aspect, fluidic components are provided. In some embodiments, the fluidic component comprises a fluidic connector comprising an inlet, the fluidic connector sized and

[0012] #14629375vl adapted to mechanically interface with a medical device, if present, a fluidic channel in fluidic communication with the inlet, a flexible sensor positioned proximate the fluidic channel and comprising one or more photodetectors and an impedance detector, the impedance detector comprising two or more electrodes, wherein the impedance detector is in electrical communication with the fluidic channel.

[0013] In some embodiments, the fluidic component comprises an inlet, a fluidic channel in fluidic communication with the inlet, a sensor positioned proximate the fluidic channel, wherein the sensor is configured to substantially simultaneously determine two or more of a cell concentration, a cell viability, and a cell type of the plurality of cells.

[0014] In yet another aspect, methods for determining a characteristic of a plurality of cells are provided. In some embodiments, the method comprises flowing a fluid comprising the plurality of cells through a fluidic channel such that the plurality of cells flows past a sensor, the sensor comprising one or more photodetectors and an impedance detector, measuring, via two or more electrodes of the impedance detector, an impedance of the fluid as it flows through the fluidic channel, thereby generating an impedance signal, exposing the fluidic channel to a source of electromagnetic radiation and detecting, via the one or more photodetectors, a change in the electromagnetic radiation, thereby generating an optical signal and determining the characteristic of the plurality of cells based upon the impedance signal and the optical signal.

[0015] In yet another aspect, reusable systems are provided. In some embodiments, the reusable system has the capacity to integrate on an inner surface of a needle, the system configured to measure one or more of cell viability, concentration, and flow speed during flow of a fluid through the needle.

[0016] Other advantages and novel features of the present disclosure will become apparent from the following detailed description of various non-limiting embodiments of the disclosure when considered in conjunction with the accompanying figures. In cases where the present specification and a document incorporated by reference include conflicting and / or inconsistent disclosure, the present specification shall control.

[0017] BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Non-limiting embodiments of the present disclosure will be described by way of example with reference to the accompanying figures, which are schematic and are not intended to be drawn to scale unless otherwise indicated. In the figures, each identical or nearly identical component illustrated is typically represented by a single numeral. For purposes of clarity, not

[0019] #14629375vl every component is labeled in every figure, nor is every component of each embodiment of the disclosure shown where illustration is not necessary to allow those of ordinary skill in the art to understand the disclosure. In the figures:

[0020] FIG. 1A shows a schematic illustration of an exemplary system, according to one set of embodiments.

[0021] FIG. IB shows a schematic illustration of an exemplary system, according to one set of embodiments.

[0022] FIG. 1C shows a schematic of a smart cell analysis needle system, according to one set of embodiments.

[0023] FIG. ID shows schematics showing the fabrication process for the smart needle system, according to one set of embodiments, a) Preparation of a flexible Cu (18 m)-PI (75 m)-Cu (18 m) tri-layer flexible printed circuit board (fPCB) substrate, b) Laser ablation process of flexible fPCB substrate forms the layout of impedance sensors as well as soldering pads for optoelectronic components and the flexible cable connector, c) Chemical vapor deposition of parylene (~ 13 pm) encapsulates the device. Followed laser ablation exposes the electrodes of the impedance sensors, d) Aligned PI shadow mask covers the fPCB substrate with only electrodes of the impedance sensors exposed, e) Electron beam vapor deposition process sequentially deposit Ti (20 nm) and Au (200 nm) on the exposed electrodes. Hot air soldering electrically bonds the p-ILEDs and p-IPDs on the corresponding soldering pads, f) Aligned attachment of Very-High-Bond (VHB) Foam Mounting on the surface of the fPCB substrate for further assembly, g-h) Zoomed in images of the impedance sensor (g) and soldered optoelectronic components (h).

[0024] FIG. 2 shows impedance characterization of the viability of human embryonic stem cells (hESC) in single cell form at static scenario, according to one set of embodiments. A) Schematic illustration of the impedance detection of cell viability inside the cell suspension from the disturbed electrical field generated and sensed by paired electrodes. B) Layered schematics of the thin film impedance sensors. C) Microscopic image of the impedance sensor with seven pairs of rectangular electrodes (1.35 x 0.05 mm2) in toothing form with 0.05 mm interspacing. D) Impedance measurement scanning from 100 to IM Hz for hESC single cells at 10 M cells / mL concentration. Top: resistive part of the impedance. Middle: capacitive part of the impedance. Bottom: absolute value of the impedance. E) The corresponding type III error statistical analysis for cell viability detection on hESC single cells at 10 M cells / mL concentration at different frequency ranges. F) Same statistical analysis as (E) for concentration detection of hESC single

[0025] #14629375vl cells at 94.3% viability. G-H) Statistical variances of the resistive part, capacitive part, and absolute values of impedance data explained by cumulative frequency predictors in respect to varying cell viability (G) and concentration (H) of hESC single cells.

[0026] FIG. 3 shows optical characterization of the concentration and viability of HEK293 cells in single cell form at static scenario, according to one set of embodiments. A) Schematics for the measurement set up for optical detection of the viability and concentration of HEK293 single cells. B) Results for optical detection on the concentration of HEK293 single cells at ~ 97% viability. C) Results for optical detection on the viability of HEK293 single cells at 1 M cells / mL concentration. D) Statistical described variances of the concentration of HEK293 single cells at ~ 97% viability across the wavelengths between 400 and 1000 nm. E-F) The corresponding type I (E) and type III (F) error statistical analysis for cell concentration detection. G-I) Same analysis as described in (D)-(F) for the detection of the viability of HEK293 single cells at IM cells / mL concentration.

[0027] FIG. 4 shows characterization of cell viability, concentration, and type in SCANS during dynamic delivery, according to one set of embodiments. A) Schematics of the integration of impedance and optical sensors on thin flexible printed circuit board (fPCB). B) Assembly process of the SCANS using the fPCB device. Scale bar: 1 cm. C) Circular sankey flow chart that explains the correlation between each statistical effect during dynamic optical sensing including the viability and concentration of cells, the wavelengths of microscale light emitting diodes (LEDs), the current and voltage changes detected by microscale photodiodes (PDs), and detection modes of paired LEDs and PDs. D-E) Classification of current and voltage changes detected by PDs at all detection modes (direct and scattered lights at both 470 nm and 525 nm) at different viabilities (D) and concentrations (E). F) Circular sankey flow chart that explains the correlation between each statistical effect during dynamic impedance sensing including the viability and concentration of cells, the frequency and repetition (scan cycle) of impedance scan, and the real part (resistive), imaginary part (capacitive), absolute values, and the phase angles of the collected impedance. G-H) Classification of impedance change (both resistive and capacitive parts) at all scanned frequencies with respect to different viabilities (G) and concentrations (H). I) Similarity heat map for classification of a total of 15 different single cells between each other using the SCANS. J) Similarity heat plots for statistical differentiation of HCT15 single cells (blue box) and BT549 single cells (brown box) from other 14 cells shown in (I).

[0028] FIG. 5 shows machine learning-assisted SCANS for augmented detection range and accuracy, according to one set of embodiments. A) Structure of random forest machine-learning

[0029] #14629375vl model for single sensor-based regression to predict cell viability and concentration. B) Structure of attention model for predicating cell viability and concentration jointly based on both optical and impedance sensors. C) Prediction accuracy defined by MAPE of the optical sensor-only model, dash box referred to the dataset for test validation. D) Prediction accuracy defined by MAPE of the impedance sensor-only model, dash box referred to the dataset for test validation. E) Prediction accuracy defined by MAPE of the attention model combining optical and impedance sensors, dash box referred to the dataset for test validation. F) Cumulative prediction accuracy when sequentially adding each optical sensor to the model (left to right), with individual MAPE of each optical sensor. 470-D: Direct light at 470 nm. 470-S: Scattered light at 470 nm. 525-D: Direct light at 525 nm. 525- S: Scattered light at 525 nm. G) Cumulative prediction accuracy when sequentially adding each impedance scanning frequency to the model, with individual MAPE of each frequency.

[0030] FIG. 6 shows a thin film sensor fabrication procedure, according to one set of embodiments. A) Fabrication of thin film sensor begins by prepare a 4-inch thermal oxide wafer sequentially cleaned by acetone, isopropyl alcohol (IPA), distilled water, and IPA rinse. B) Spin coating process forms a ~ 2 pm thick film of polyimide (PI) on the wafer. C) Electron beam vapor (E-Beam) deposition formed metallic layers of Cr (5 nm) and Au (200 nm) sequentially on the PI layer. D) Photolithographical patterning followed by wet etching process defines the layouts of the impedance sensors and the contact pads. E) Spin coating process forms a second ~ 2 pm thick PI layer on top of the patterned metallic layers. F) Photolithographical patterning followed by thermal evaporation and acetone lift off of Ni (60 nm) forms a hard mask that covers the PI with electrodes of the impedance sensors and contact pads for electrical connections exposed. G) Reactive ion etching removes the polyimide on top of the electrodes and contact pads. Then the Ni etchant removes the Ni hard mask. H) Zoomed in view of the exposed electrodes and contact pads. I) A water-soluble tape transfers the devices from the wafer substrate after cutting the wafer in half. J-L) Detailed illustrations of the transfer process. First, attach the water-soluble tape on the device side of the wafer (J). Then, flip and cut the wafer in half using a diamond cutter (K). Finally, detach the wafer from the tape as depicted in (L).

[0031] FIG. 7 shows a thin film impedance sensor designs, according to one set of embodiments. A) Autocad design for impedance sensor with 0.5 x 0.51 mm2lateral dimension with 13 pairs of rectangular electrodes (0.5 x 0.01 mm2) with 0.01 mm interspacing. B) Autocad design for impedance sensor with 1.0 x 0.98 mm2lateral dimension with 25 pairs of rectangular electrodes (1.0 x 0.01 mm2) with 0.01 mm interspacing. C) Autocad design for impedance sensor with 1.35

[0032] #14629375vl x 1.35 mm2lateral dimension with 7 pairs of rectangular electrodes (1.35 x 0.05 mm2) with 0.05 mm interspacing.

[0033] FIG. 8 shows impedance detection of the viability of human embryonic stem cells (hESC) in single cell form at static scenario using 0.5 mm size impedance sensor, according to one set of embodiments. A) Resistive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells with different viabilities at 1 M cells / mL concentration. The impedance sensor (0.5 x 0.51 mm2) has 13 pairs of rectangular electrodes (0.5 x 0.01 mm2) with 0.01 mm interspacing (FIG. 7). B) Resistive part of the impedance at 1 kHz (marked in pink in (A)) with respect to the viability. C) Resistive part of the impedance at 10 kHz (marked in blue in (A)) with respect to the viability. d-F) Repeat (A)-(C) for the capacitive part of the impedance scan. G-L) Repeat (A)-(F) for 10 pL hESC single cells with different viabilities at 10 M cells / mL concentration.

[0034] FIG. 9 shows impedance detection (resistive part) of the concentration of human embryonic stem cells (hESC) in single cell form at static scenario using 0.5 mm size impedance sensor, according to one set of embodiments. A) Resistive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells at -78.0% viability with different concentrations (10, 20, 40, and 80 k cells / mL). The impedance sensor (0.5 x 0.51 mm2) has 13 pairs of rectangular electrodes (0.5 x 0.01 mm2) with 0.01 mm interspacing (FIG. 7). B) Repeat (A) with concentrations at 100, 200, 400, and 800 k cells / mL. C) Repeat (A) with concentrations at 1, 2, 4, 8, and 10 M cells / mL. D) Resistive part of the impedance at 1 kHz (marked in pink in (A)-(C)) with respect to the concentration. E) Resistive part of the impedance at 10 kHz (marked in blue in (A)-(C)) with respect to the concentration. F) Resistive part of the impedance at 1 MHz (marked in brown in (A)-(C)) with respect to the concentration.

[0035] FIG. 10 shows impedance detection (capacitive part) of the concentration of human embryonic stem cells (hESC) in single cell form at static scenario using 0.5 mm size impedance sensor, according to one set of embodiments. A) Capacitive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells at -78.0% viability with different concentrations (10, 20, 40, and 80 k cells / mL). The impedance sensor (0.5 x 0.51 mm2) has 13 pairs of rectangular electrodes (0.5 x 0.01 mm2) with 0.01 mm interspacing (FIG. 7). B) Repeat (A) with concentrations at 100, 200, 400, and 800 k cells / mL. C) Repeat (A) with concentrations at 1, 2, 4, 8, and 10 M cells / mL. D) Capacitive part of the impedance at 1 kHz (marked in pink in (A)-(C)) with respect to the concentration. E) Capacitive part of the

[0036] #14629375vl impedance at 10 kHz (marked in blue in (A)-(C)) with respect to the concentration. F) Capacitive part of the impedance at 1 MHz (marked in brown in (A)-(C)) with respect to the concentration.

[0037] FIG. 11 shows impedance detection of the viability of human embryonic stem cells (hESC) in single cell form at static scenario using 1 mm size impedance sensor, according to one set of embodiments. A) Resistive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells with different viabilities at 1 M cells / mL concentration. The impedance sensor (1.0 x 0.98 mm2) has 25 pairs of rectangular electrodes (1.0 x 0.01 mm2) with 0.01 mm interspacing (FIG. 7). B) Resistive part of the impedance at 1 kHz (marked in pink in (A)) with respect to the viability. C) Resistive part of the impedance at 10 kHz (marked in blue in (A)) with respect to the viability. D-F) Repeat (A)-(C) for the capacitive part of the impedance scan. G-L) Repeat (A)-(F) for 10 pL hESC single cells with different viabilities at 10 M cells / mL concentration.

[0038] FIG. 12 shows impedance detection (resistive part) of the concentration of human embryonic stem cells (hESC) in single cell form at static scenario using 1 mm size impedance sensor, according to one set of embodiments. A) Resistive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells at -78.0% viability with different concentrations (10, 20, 40, and 80 k cells / mL). The impedance sensor (1.0 x 0.98 mm2) has 25 pairs of rectangular electrodes (1.0 x 0.01 mm2) with 0.01 mm interspacing (FIG. 7). B) Repeat (A) with concentrations at 100, 200, 400, and 800 k cells / mL. C) Repeat (A) with concentrations at 1, 2, 4, 8, and 10 M cells / mL. D) Resistive part of the impedance at 1 kHz (marked in pink in (A)-(C)) with respect to the concentration. E) Resistive part of the impedance at 10 kHz (marked in blue in (A)-(C)) with respect to the concentration. F) Resistive part of the impedance at 1 MHz (marked in brown in (A)-(C)) with respect to the concentration.

[0039] FIG. 13 shows impedance detection (capacitive part) of the concentration of human embryonic stem cells (hESC) in single cell form at static scenario using 1 mm size impedance sensor, according to one set of embodiments. A) Capacitive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells at -78.0% viability with different concentrations (10, 20, 40, and 80 k cells / mL). The impedance sensor (1.0 x 0.98 mm2) has 25 pairs of rectangular electrodes (1.0 x 0.01 mm2) with 0.01 mm interspacing (FIG. 7). B) Repeat (A) with concentrations at 100, 200, 400, and 800 k cells / mL. C) Repeat (A) with concentrations at 1, 2, 4, 8, and 10 M cells / mL. D) Capacitive part of the impedance at 1 kHz (marked in pink in (A)-(C)) with respect to the concentration. E) Capacitive part of the

[0040] #14629375vl impedance at 10 kHz (marked in blue in (A)-(C)) with respect to the concentration. F) Capacitive part of the impedance at 1 MHz (marked in brown in (A)-(C)) with respect to the concentration.

[0041] FIG. 14 shows impedance detection of the viability of human embryonic stem cells (hESC) in single cell form at static scenario using 1.35 mm size impedance sensor, according to one set of embodiments. A) Resistive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells with different viabilities at 1 M cells / mL concentration. The impedance sensor (1.35 x 1.35 mm2) has 7 pairs of rectangular electrodes (1.35 x 0.05 mm2) with 0.05 mm interspacing (FIG. 7). B) Resistive part of the impedance at 1 kHz (marked in pink in (A)) with respect to the viability. C) Resistive part of the impedance at 10 kHz (marked in blue in (A)) with respect to the viability. D-F) Repeat (A)-(C) for the capacitive part of the impedance scan. G-L) Repeat (A)-(F) for 10 pL hESC single cells with different viabilities at 10 M cells / mL concentration.

[0042] FIG. 15 shows impedance detection (resistive part) of the concentration of human embryonic stem cells (hESC) in single cell form at static scenario using 1.35 mm size impedance sensor, according to one set of embodiments. A) Resistive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells at -78.0% viability with different concentrations (10, 20, 40, and 80 k cells / mL). The impedance sensor (1.35 x 1.35 mm2) has 7 pairs of rectangular electrodes (1.35 x 0.05 mm2) with 0.05 mm interspacing (FIG. 7). B) Repeat (A) with concentrations at 100, 200, 400, and 800 k cells / mL. C) Repeat (A) with concentrations at 1, 2, 4, 8, and 10 M cells / mL. D) Resistive part of the impedance at 1 kHz (marked in pink in (A)-(C)) with respect to the concentration. E) Resistive part of the impedance at 10 kHz (marked in blue in (A)-(C)) with respect to the concentration. F) Resistive part of the impedance at 1 MHz (marked in brown in (A)-(C)) with respect to the concentration.

[0043] FIG. 16 shows impedance detection (capacitive part) of the concentration of human embryonic stem cells (hESC) in single cell form at static scenario using 1.35 mm size impedance sensor, according to one set of embodiments. A) Capacitive part of the impedance scanned between 0.1 and 1000 kHz in logarithmic scale for 10 pL hESC single cells at -78.0% viability with different concentrations (10, 20, 40, and 80 k cells / mL). The impedance sensor (1.35 x 1.35 mm2) has 7 pairs of rectangular electrodes (1.35 x 0.05 mm2) with 0.05 mm interspacing (FIG. 7). B) Repeat (A) with concentrations at 100, 200, 400, and 800 k cells / mL. C) Repeat (A) with concentrations at 1, 2, 4, 8, and 10 M cells / mL. D) Capacitive part of the impedance at 1 kHz (marked in pink in (A)-(C)) with respect to the concentration. E) Capacitive part of the

[0044] #14629375vl impedance at 10 kHz (marked in blue in (A)-(C)) with respect to the concentration. F) Capacitive part of the impedance at 1 MHz (marked in brown in (A)-(C)) with respect to the concentration.

[0045] FIG. 17 shows impedance and optical sensors of SCANS, according to one set of embodiments. A) Zoomed in image of a Ti / Au coated impedance sensor on the fPCB. B) Zoomed in image of the soldered pLED and pPD.

[0046] FIG. 18 shows different detection mode enabled by the optical sensing system of the SCANS, according to one set of embodiments. The four independent detection modes include the collection of optical signals from photodiode 1 which collects direct light at 470 nm (Direct blue) and scattered light at 525 nm (Scattered green) as well as the collection of optical signals from photodiode 2 which collects direct light at 525 nm (Direct green) and scattered light at 470 nm (Scattered blue).

[0047] FIG. 19 shows assembly of the SCANS, according to one set of embodiments. A) Flexible printed circuit boards (fPCB) with soldered microscale light emitting diodes (pLEDs) and photodiodes (pPDs). B) Laser ablated very-high-bond (VHB) foam mounting tape adheres on the surface of the fPCB in alignment. C) The inner surface of the sensing module made from shrink wrap on an assembly rod. The inset image shows the hole made into the shrink wrap before placing it back onto the assembly rod. D) The fPCB next to the assembly rod with the shrink wrap, indicating the necessity of alignment between the circular holes on the shrink wrap and the circular holes in the tape for both impedance sensors. E) The rolled fPCB fully encased with the outer shrink wrap. F) The sensing module taking off from the assembly rod with the connection pads for external hardware connections pulled out of the outer shrink wrap. A commercial luer lock adapter is cut in half with the sensing module placed in between. G) Three 1 ml syringes connected with fully assembled sensing modules to form the SCANS. H) An example of the compatibility of the sensing module with the commercial interchangeable delivery needles with different gauge sizes. I) Devices with assembled SCANS. Each device contains a soldered 12-pin flat flexible cable (FFC) connector that transmits the collected signals to the external hardware for analysis. Bottom - phosphate buffered saline (PBS) loaded into the syringe equipped with the SCANS for delivery; Top - PBS with 60 million human embryonic stem cells (hESCs; single cell form) at 79.4% viability loaded into an identical device, giving a semi-opaque appearance.

[0048] FIG. 20 shows operation of the SCANS during dynamic delivery of cells, according to one set of embodiments. A) Experimental set up of a syringe pump that forms a continuous flow of cell suspension with programmed speed through the SCANS and transmits collected

[0049] #14629375vl impedance and optical signals to external analytical hardware and software via a flat flexible cable and a customized connector board. A customized connector board connect the SCANS through flat flexible cable. The board transmit the collected signals to external hardware and processing software. B) Zoomed in view of the fixation of the syringe loaded on a syringe pump. C) Impedance baseline measurement of phosphate buffered saline (PBS) at 1, 5, 10, 100, and 1000 kHz after random positioning of the device. D) Optic baseline measurement of PBS from four pairs of microscale light emitting diodes and photodiodes after random positioning of the device.

[0050] FIG. 21 shows assembly process and prototype for the needle version of the SCANS, according to one set of embodiments. A-B) Schematics (A) and images (B) of the assembly process of integrating thin-film impedance sensors into the commercial delivery needle.

[0051] FIG. 22 shows control of cell viability in cluster form, according to one set of embodiments. A) Morphologies of INS- 1 cell clusters at different viabilities. Scale bar: 100 micrometers B) Control of the viability of INS- 1 cell clusters using different concentrations of dimethyl sulfoxide (DMSO) solution.

[0052] FIG. 23 morphology of different cell clusters at different viabilities, according to one set of embodiments. A) Microscopic images of BC09 cell clusters at viabilities of ~ 71%, ~ 65%, and ~ 8.6%. B) Microscopic images of human embryonic stem cell (hESC) clusters at viabilities of ~ 95%, ~ 80%, ~ 65%, and ~ 30%. C) Microscopic images of cardiomyocyte clusters at viabilities of ~ 80% and ~ 40%.

[0053] FIG. 24 shows detection of the concentration of cell clusters using the optical sensors in the SCANS during dynamic delivery, according to one set of embodiments. A) The current change with respect to the concentration (2, 4, 6, 8, and 10 M cells / mL) of human embryonic stem cells (hESC) in cluster form at 98.7% viability from the photodiode that collects the direct light signal at 470 nm (i), scattered light signal at 470 nm (ii), direct light signal at 525 nm (iii), and scattered light signal at 525 nm (iv) that penetrate through the cell suspension in the SCANS. B) The current change with respect to the concentration (2, 4, 6, 8, 10, 20, and 30 M cells / mL) of BC09 cells (hESC) in cluster form at 78.8% viability from the photodiode that collects the direct light signal at 470 nm (i), scattered light signal at 470 nm (ii), direct light signal at 525 nm (iii), and scattered light signal at 525 nm (iv) that penetrate through the cell suspension in the SCANS.

[0054] FIG. 25 shows performance of the SCANS in dynamic delivery of human embryonic stem cells (hESC) in single cell form using the equipped optical sensors, according to one set of embodiments. A) A heatmap of the current change with respect to both the concentration (2, 6,

[0055] #14629375vl 10, and 20 M cells / mL) and viability (70.1%, 79.6%, 90.0%, and 97.4%) of hESC single cells from the photodiode that collects the direct light signal at 470 nm that penetrates through the cell suspension in the SCANS. B) The detected current change with respect to the concentration of hESC single cells at 97.4% viability, which is marked in the orange arrow in (A). C) The detected current change with respect to the viability of the hESC single cells at 6 M cells / mL concentration, which is marked in the purple line in (A). D-F) Repeat (A)-(C) with the photodiode that collects the scattered light signal at 470 nm. G-I) Repeat (A)-(C) with the photodiode that collects the direct light signal at 525 nm. J-L) Repeat (A)-(C) with the photodiode that collects the scattered light signals at 525 nm. All four light signal collections occur simultaneously during the delivery of cell suspension.

[0056] FIG. 26 shows performance of the SCANS in dynamic delivery of human embryonic stem cells (hESC) in cluster form using the equipped optical sensors at low concentration, according to one set of embodiments. A) A heatmap of the current change with respect to both the concentration (6, 10, and 20 M cells / mL) and viability (64.8%, 79.4%, and 98.7%) of hESC clusters from the photodiode that collects the direct light signal at 470 nm that penetrates through the cell suspension in the SCANS. B) The detected current change with respect to the concentration of hESC clusters at 79.4% viability, which is marked in the orange arrow in (A). C) The detected current change with respect to the viability of the hESC clusters at 10 M cells / mL concentration, which is marked in the purple line in (A). D-F) Repeat (A)-(C) with the photodiode that collects the scattered light signal at 470 nm. G-I) Repeat (A)-(C) with the photodiode that collects the direct light signal at 525 nm. J-L) Repeat (A)-(C) with the photodiode that collects the scattered light signals at 525 nm. All four light signal collections occur simultaneously during the delivery of cell suspension.

[0057] FIG. 27 shows performance of the SCANS in dynamic delivery of human embryonic stem cells (hESC) in cluster form using the equipped optical sensors at high concentration, according to one set of embodiments. A) A heatmap of the current change with respect to both the concentration (40 and 60 M cells / mL) and viability (64.8%, 79.4%, and 98.7%) of hESC clusters from the photodiode that collects the direct light signal at 470 nm that penetrates through the cell suspension in the SCANS. B) The detected current change with respect to the concentration of hESC clusters at 98.7% viability, which is marked in the orange arrow in (A). C) The detected current change with respect to the viability of the hESC clusters at 40 M cells / mL concentration, which is marked in the purple line in (A). D-F) Repeat (A)-(C) with the photodiode that collects the scattered light signal at 470 nm. G-I) Repeat (A)-(C) with the photodiode that collects the

[0058] #14629375vl direct light signal at 525 nm. J-L) Repeat (A)-(C) with the photodiode that collects the scattered light signals at 525 nm. All four light signal collections occur simultaneously during the delivery of cell suspension.

[0059] FIG. 28 shows performance of the SCANS in dynamic delivery of cardiomyocytes in cluster form using the equipped optical sensors, according to one set of embodiments. A) A heatmap of the current change with respect to both the concentration (4 and 14 M cells / mL) and viability (44.9%, and 85.1%) of cardiomyocyte clusters from the photodiode that collects the direct light signal at 470 nm that penetrates through the cell suspension in the SCANS. B) The detected current change with respect to the concentration of cardiomyocyte clusters at 44.9% viability, which is marked in the orange arrow in (A). C) The detected current change with respect to the viability of the cardiomyocyte clusters at 14 M cells / mL concentration, which is marked in the purple line in (A). D-F) Repeat (A)-(C) with the photodiode that collects the scattered light signal at 470 nm. G-I) Repeat (A)-(C) with the photodiode that collects the direct light signal at 525 nm. J-L) Repeat (A)-(C) with the photodiode that collects the scattered light signals at 525 nm. All four light signal collections occur simultaneously during the delivery of cell suspension.

[0060] FIG. 29 shows performance of the SCANS in dynamic delivery of human embryonic stem cell derived beta cells (BC09) in cluster form using the equipped optical sensors, according to one set of embodiments. A) A heatmap of the current change with respect to both the concentration (6, 10, and 20 M cells / mL) and viability (4.5%, and 78.8%) of BC09 clusters from the photodiode that collects the direct light signal at 470 nm that penetrates through the cell suspension in the SCANS. B) The detected current change with respect to the concentration of BC09 clusters at 78.8% viability, which is marked in the orange arrow in (A). C) The detected current change with respect to the viability of the BC09 clusters at 20 M cells / mL concentration, which is marked in the purple line in (A). D-F) Repeat (A)-(C) with the photodiode that collects the scattered light signal at 470 nm. G-I) Repeat (A)-(C) with the photodiode that collects the direct light signal at 525 nm. J-L) Repeat (A)-(C) with the photodiode that collects the scattered light signals at 525 nm. All four light signal collections occur simultaneously during the delivery of cell suspension.

[0061] FIG. 30 shows characterization of cell viability and concentration of human embryonic stem cells (hESC) in cluster form in SCANS during dynamic delivery, according to one set of embodiments. A) Circular Sankey flow chart that explains the correlation between each statistical effect during dynamic optical sensing including the viability and concentration of cells, the

[0062] #14629375vl wavelengths of microscale light emitting diodes (pLEDs), the current and voltage changes detected by microscale photodiodes (pPDs), and detection modes of paired pLEDs and pPDs. B- C) Classification of current and voltage changes detected by pPDs at all detection modes (direct and scattered lights at 535 nm) at different viabilities (B) and concentrations (C). D) Circular Sankey flow chart that explains the correlation between each statistical effect during dynamic impedance sensing including the viability and concentration of cells, the frequency and repetition (scan cycle) of impedance scan, and the real part (resistive), imaginary part (capacitive), absolute values, and the phase angles of the collected impedance. E-F) Classification of impedance change (both resistive and capacitive parts) at all scanned frequencies with respect to different viabilities (E) and concentration (F).

[0063] FIG. 31 shows characterization of cell viability and concentration of human embryonic stem cell derived beta cells (BC09) in cluster form in SCANS during dynamic delivery, according to one set of embodiments. A) Circular Sankey flow chart that explains the correlation between each statistical effect during dynamic optical sensing including the viability and concentration of cells, the wavelengths of microscale light emitting diodes (pLEDs), the current and voltage changes detected by microscale photodiodes (pPDs), and detection modes of paired pLEDs and pPDs. B-C) Classification of current and voltage changes detected by pPDs at all detection modes (direct and scattered lights at 535 nm) at different viabilities (B) and concentrations (C). D) Circular Sankey flow chart that explains the correlation between each statistical effect during dynamic impedance sensing including the viability and concentration of cells, the frequency and repetition (scan cycle) of impedance scan, and the real part (resistive), imaginary part (capacitive), absolute values, and the phase angles of the collected impedance. E- F) Classification of impedance change (both resistive and capacitive parts) at all scanned frequencies with respect to different viabilities (E) and concentration (F).

[0064] FIG. 32 shows characterization of cell viability and concentration of cardiomyocytes in cluster form in SCANS during dynamic delivery, according to one set of embodiments. A) Circular Sankey flow chart that explains the correlation between each statistical effect during dynamic optical sensing including the viability and concentration of cells, the wavelengths of microscale light emitting diodes (pLEDs), the current and voltage changes detected by microscale photodiodes (pPDs), and detection modes of paired pLEDs and pPDs. B-C) Classification of current and voltage changes detected by pPDs at all detection modes (direct and scattered lights at 535 nm) at different viabilities (B) and concentrations (C). D) Circular Sankey

[0065] #14629375vl flow chart that explains the correlation between each statistical effect during dynamic impedance sensing including the viability and concentration of cells, the frequency and repetition (scan cycle) of impedance scan, and the real part (resistive), imaginary part (capacitive), absolute values, and the phase angles of the collected impedance. E-F) Classification of impedance change (both resistive and capacitive parts) at all scanned frequencies with respect to different viabilities (E) and concentration (F).

[0066] FIG. 33 shows distinguishing cell types by impedance and optical sensing, according to one set of embodiments. A-C) Heat map of similarities among 15 different cell types from statistical models using data from impedance sensing (A), optical sensing (B), and their combinations (C).

[0067] FIG. 34 shows machine learning model for prediction of cell concentration and viability, according to one set of embodiments. Structure of a random forest model for single-sensor-based regression to predict cell concentration and viability.

[0068] FIG. 35 shows different train-split models for predicting the viability and concentration of hESC single cells, according to one set of embodiments. Increment of training data leads to significant improvement of predication accuracy defined by MAPE of the attention model in distinguishing cell types by impedance and optical sensing, comparing with the original 9:7 traintest split strategy (FIG. 5D). A) and B) Adding only one data pair of the unknown concentration 20 M cells / mL and the known viability of 70.1% leads to significant improvement of prediction accuracy in training dataset. C) and D) Adding only one data pair of both the unknown concentration 20 M cells / mL and the unknown viability of 90% leads to significant improvement in prediction of training dataset by 10 folds, and in most prediction of test data by 2 folds approximately.

[0069] FIG. 36 shows different train-split models for predicting the viability and concentration of hESC clusters, according to one set of embodiments. Distinguishing cell types of hESC cell clusters by impedance and optical sensing with the joint attention model (dash redline represents the test dataset). A) and B) Prediction MAPE (%) with a 4-2 train-test split yields high-accuracy prediction results. C) and D) Increasing the train-test split to 5-1 significantly improves the prediction accuracy.

[0070] FIG. 37 shows different train- split models for predicting the viability and concentration of BC09 clusters, according to one set of embodiments. Distinguishing cell types of BC09 cell clusters by impedance and optical sensing with the joint attention model (dash redline represents the test dataset). A) and B) Prediction MAPE (%) with a 3-3 aggressive train-test split. C) and D)

[0071] #14629375vl Increasing the train-test split to 5-1 yields near-perfect prediction accuracy, with MAPE mostly below 5% across all train and test data.

[0072] FIG. 38 shows different train- split models for predicting the viability and concentration of cardiomyocyte clusters, according to one set of embodiments. Distinguishing cell types of Cardiomyocyte cell clusters by impedance and optical sensing with the joint attention model (dash redline represents the test dataset). A) and B) Prediction MAPE (%) with a 2-2 aggressive train-test split. C) and D) Increasing the train-test split to 3-1 significantly improves the prediction accuracy on viability. The limited improvement in prediction test concentration data is understandable due to the extremely limited amount of 3 pairs of training data.

[0073] DETAILED DESCRIPTION

[0074] Smart needle systems and related methods, such as smart injection needle system for accurate characterization of cell types, concentrations, and viability, are generally provided. In some embodiments, fluidic components (e.g., configured to interface with a medical device such as a syringe or in-line flow system) are provided. Accurate assessment of cell therapies including precise evaluation of cell type and form (single cell or cluster), viability, and concentration is critical for effective medical treatment both in preparing the therapy and at the point-of-care. Current mainstream technologies for cell characterization rely on electrode-based or fluorescence-based method with large, heavy, and bulky equipment. Application of such equipment require sampling which only provides an average estimate of total cell population. Furthermore, the measurements from such equipment generally cannot reflect the conditions of the cells at the point-of-care right before therapeutic administration, which could be largely different than the characteristics during preparation. Advantageously, the articles, systems, and methods described herein may be, in some embodiments, conveniently deployed in-line between e.g., commercial syringes and needles, and / or directly into the needles themselves. In some embodiments, the smart needle system combines advanced micro-fabrication processes to achieve multi-mode sensing in a miniaturized form factor, with machine learning (ML) processes, to offer e.g., real-time rapid analysis of cell populations at the time and location of medical administration.

[0075] In some embodiments, the system comprises a fluidic component. In some embodiments, the fluidic component comprises a fluidic channel configured to receive a fluid comprising a plurality of particles, such a plurality of cells (e.g., a plurality of cells obtained from a subject). The term “subject," as used herein, refers to an individual organism such as a human or an

[0076] #14629375vl animal. In some embodiments, the subject is a mammal (e.g., a human, a non-human primate, or a non-human mammal), a vertebrate, a laboratory animal, a domesticated animal, an agricultural animal, or a companion animal. In some embodiments, the subject is a human. In some embodiments, the subject is a rodent, a mouse, a rat, a hamster, a rabbit, a dog, a cat, a cow, a goat, a sheep, or a pig.

[0077] In some embodiments, the system may be used to determine one or more characteristics of a particle (e.g., a cell). Non-limiting examples of characteristics include particle size, particle concentration, particle type. For example, in the context of cells, the one or more characteristics may be selected from the group consisting of cell viability, cell concentration, and cell type. In some embodiments, the system is capable of determining two or more characteristics in a single use (e.g., cell viability and cell type, cell concentration and cell type, cell viability and cell concentration, cell viability, cell concentration, and cell type). In some embodiments, the system may be used to determine a flow rate of the cells. Other characteristics are also possible. In some embodiments, the system is capable of determining two or more characteristics of a cell substantially simultaneously.

[0078] In some embodiments, the fluidic component comprises a sensor. In some embodiments, the sensor comprises two or more components (e.g., an impedance detector, an optical component). In some embodiments, the sensor is flexible (e.g., deposited on a flexible printed circuit board (fPCB)). In an exemplary set of embodiments, the sensor comprises a trilayer (e.g., metal-polymer-metal) structure. Non-limiting examples of suitable metals include gold, titanium, chromium, zinc, iron, copper, nickel, and tin. In some embodiments, the metal is a metal oxide semiconductor. Non-limiting examples of suitable metal oxide semiconductors include zinc oxide, copper oxide, titanium dioxide, tin oxide, aluminum oxide, and indium oxide. Other metals and metal oxide semiconductors are also possible. In an exemplary set of embodiments, the metal is copper.

[0079] Non-limiting examples of suitable polymers include poly dimethylsiloxane, polyethylene terephthalate, poly(ether-ether-ketone), polycarbonate, polyvinyl chloride, polyethylene naphthalate, polyurethane, polyvinylidene fluoride, and polyimide. In an exemplary set of embodiments, the polymer is polyimide. Other polymers are also possible.

[0080] In some embodiments, one or more components of the sensor (e.g., electrodes) are deposited lithographically on the substrate.

[0081] In some embodiments, the fluidic component comprises a fluidic channel. In some embodiments, the fluidic channel is configured to receive a fluid (e.g., comprising a plurality of

[0082] #14629375vl cells). In an exemplary set of embodiments, the fluidic channel is in a fluidic component configured to mechanically engage with a syringe and / or a needle. For example, in some embodiments, the fluidic component comprises a luer connector (e.g., luer lock, luer slip, and combinations thereof) comprising the fluidic channel. Other connectors are also possible (e.g., pressure fit couplings, standard couplings, or the like). In some embodiments, the connection may be provided by a non-threaded, non-Luer-type fitting, including a friction fit, interference fit, pressure fit, compression fit, or barbed fitting, each of which may provide retention and, in some cases, a fluid-tight seal.

[0083] As shown in an exemplary embodiment in FIG. 1A, system 100 comprises fluidic channel 110 (e.g., a fluidic channel configured to receive a fluid comprising a plurality of cells) and sensor 120 associated with fluidic channel 110. In some embodiments, sensor 120 is adjacent fluidic channel 110. In some embodiments, sensor 120 is directly adjacent fluidic channel 110. As used herein, when a component is referred to as being “adjacent” another component, it can be directly adjacent to (e.g., in contact with) the component, or one or more intervening components (e.g., polymeric materials, layers, coatings) may also be present. A component that is “directly adjacent” another component means that no intervening component(s) is present. Sensor 120 may comprise: an impedance detector (not shown) comprising two or more electrodes in electrical communication with the fluidic channel 110; and an optical component (not shown) proximate the fluidic channel 110, wherein the optical component comprises one or more photodetectors. Sensor 120 may be configured to determine one or more characteristics of the plurality of cells based upon a signal from the impendence detector and / or the one or more photodetectors as the fluid flows through the fluidic channel 110. The one or more characteristics may be selected from the group consisting of cell viability, cell concentration, and / or cell type. The optical component may comprise a source of electromagnetic radiation.

[0084] In some embodiments, system 100 comprises a connector 130 (e.g., a luer connector) configured and adapted to mechanically interface with a medical device component (e.g., a syringe, a needle, a catheter, a fluidic tube). Connector 130 is associated with sensor 120. In some embodiments, connector 130 is adjacent sensor 120. In some embodiments, connector 130 is directly adjacent sensor 120.

[0085] In some embodiments, as shown illustratively in FIG. IB, sensor 120 comprises one or more of optical component 140 proximate fluidic channel 110 and comprising one or more photodetectors, impedance detector 150 comprising two or more electrodes in electrical communication with fluidic channel 110, and a source of electromagnetic radiation 160. Other

[0086] #14629375vl electronic components and / or sensors are also possible (e.g., a temperature sensor, a pressure sensor, an oxygen sensor, a carbon dioxide sensor, a pH sensor, a humidity sensor, a flow (e.g., flow rate) sensor etc.). In some embodiments, the sensor is configured to operate at least partially exposed to a liquid, e.g., under physiological conditions. In some embodiments, the sensor is configured to operate under ambient conditions (e.g., room temperature and atmospheric pressure).

[0087] In some embodiments, the impedance detector is positioned proximate the fluidic channel (e.g., fluidic channel 110 of FIG. 1A). In some embodiments, the impedance detector is in fluidic communication with the fluidic channel. In some embodiments, the impedance detector comprises two or more electrodes. For example, in some embodiments, the impedance detector comprising two pairs of electrodes in a toothing geometry (see e.g., FIG. 2C). In some embodiments, each pair comprises a plurality of electrodes. In some embodiments, each pair comprises one or more electrodes, two or more electrodes, three or more electrodes, four or more electrodes, five or more electrodes, six or more electrodes, seven or more electrodes, eight or more electrodes, ten or more electrodes, twenty or more electrodes, or 30 or more electrodes. Each pair of electrodes may comprise any suitable interspacing. In some embodiments, the interspacing between electrodes is greater than or equal to 10 microns, greater than or equal to 20 microns, greater than or equal to 40 microns, greater than or equal to 60 microns, greater than or equal to 80 microns, greater than or equal to 100 microns, greater than or equal to 120 microns, greater than or equal to 140 microns, greater than or equal to 160 microns, or greater than or equal to 180 microns. In some embodiments, the interspacing between electrodes is less than or equal to 200 microns, less than or equal to 180 microns, less than or equal to 160 microns, less than or equal to 140 microns, less than or equal to 120 microns, less than or equal to 100 microns, less than or equal to 80 microns, lesson or equal to 60 microns, less or equal to 40 microns, or less than or equal to 20 microns. Combinations of the above reference ranges are also possible (e.g., greater than or equal to 10 microns and less than or equal to 200 microns). Other ranges are also possible.

[0088] The impedance detector (e.g., and / or the electrodes) may have any suitable thickness. For example, in some embodiments, the thickness of the impedance detector is less than or equal to 500 nanometers, less than or equal to 400 nanometers, less than or equal to 300 nanometers, less than or equal to 200 nanometers, less than or equal to 100 nanometers, less than or equal to 50 nanometers, less than or equal to 25 nanometers. In some embodiments, the thickness of the impedance detector is greater than or equal to 10 nanometers, greater than or equal to 25

[0089] #14629375vl nanometers, greater than or equal to 50 nanometers, greater than or equal to 100 nanometers, greater than or equal to 200 nanometers, greater than or equal to 300 nanometers, or greater than or equal to 400 nanometers. Combinations of reference ranges are also possible (e.g., less than or equal to 500 nanometers and greater than or equal to 10 nanometers, less than or equal to 200 nanometers and greater than or equal to 10 nanometers). Other ranges are also possible.

[0090] In some embodiments, the impedance detector is configured to be operated at a particular frequency. In some embodiments, the frequency is greater than or equal to 100 Hz, greater than or equal to 1 kHz, greater than or equal to 2 kHz, greater than or equal to 5 kHz, greater than or equal to 10 kHz, greater than or equal to 20 kHz, greater than or equal to 50 kHz, greater than or equal to 75 kHz, or greater than or equal to 90 kHz. In some embodiments, the frequency is less than or equal to 1000 kHz, less than or equal to 500 kHz, 100 kHz, less than or equal to 90 kHz, less than or equal to 75 kHz, less than or equal to 50 kHz, less than or equal to 20 kHz, less than or equal to 15 kHz, less than or equal to 10 kHz, less than or equal to five kHz, or less than or equal to two kHz. Combinations of the above reference ranges are also possible (e.g., greater than or equal to 1 kHz and less than or equal to 100 kHz, greater than or equal to 100 Hz and less than or equal to 1000 kHz). Other ranges are also possible.

[0091] In some embodiments, the impedance detector is deposited on a flexible substrate (e.g., a fPCB, a flexible polymer substrate). In some embodiments, the flexible substrate is transparent (e.g., to a particular wavelength(s) of light detectable by the optical component and / or emitted by the source of electromagnetic radiation). In some embodiments, the impedance detector is positioned proximate the fluidic channel. In some embodiments, the impedance detector is adjacent (e.g., directly adjacent) the fluidic channel. In some embodiments, the impedance detector is embedded within a polymer. Non-limiting examples of suitable polymers include polydimethylsiloxane, polyethylene terephthalate, poly(ether-ether-ketone), polycarbonate, polyvinyl chloride, polyethylene naphthalate, polyurethane, polyvinylidene fluoride, and polyimide. In an exemplary set of embodiments, the polymer is polyimide.

[0092] In some embodiments, a system comprises an optical component. An optical component is generally configured to detect electromagnetic radiation and to output signals (e.g., electrical signals) that may be used to determine a characteristic (e.g., of a particle such as a cell). Any suitable type of optical component may be used to detect an emission (or absence of an emission. Non-limiting examples of suitable optical component include complementary metal oxide semiconductor (CMOS) sensors, charge-coupled device (CCD) sensors, and photodiodes. Those of ordinary skill in the art would be capable of selecting suitable optical component based upon

[0093] #14629375vl the teachings of this specification. The optical component may be configured, in some embodiments, with an accompanying source of electromagnetic radiation. In some embodiments, the source of electromagnetic radiation emits light having a range between at least 350 nm and less than or equal to 800 nm and has a peak in the range of spanning at least 350 nm, at least 360 nm, at least 370 nm, at least 380 nm, at least 390 nm, at least 400 nm, at least 500 nm, at least 600 nm, at least 700 nm, or at least 800 nm. In certain instances, the source of electromagnetic radiation is configured to emit electromagnetic radiation in a wavelength range of greater than or equal to 350 nm, greater than or equal to 400 nm, greater than or equal to 450 nm, greater than or equal to 500 nm, greater than or equal to 550 nm, greater than or equal to 600 nm, greater than or equal to 650 nm, greater than or equal to 700 nm, or greater than or equal to 750 nm and less than or equal to 800 nm, less than or equal to 750 nm, less than or equal to 700 nm, less than or equal to 650 nm, less than or equal to 600 nm, less than or equal to 550 nm, less than or equal to 500 nm, less than or equal to 450 nm, or less than or equal to 400 nm. Combinations of the abovereferenced ranges are also possible (e.g., greater than or equal to 350 nm and less than or equal to 800 nm). Other ranges are also possible.

[0094] In some embodiments, the optical component comprises two or more sources of electromagnetic radiation. In some embodiments, each source of electromagnetic radiation may emit the same or different wavelengths of electromagnetic radiation.

[0095] Non-limiting examples of suitable sources of electromagnetic radiation include, but are not limited to, light-emitting diodes (LEDs), organic light-emitting diodes (OLEDs), flash bulbs, emissive species (e.g., fluorescent dyes, inorganic phosphors), ambient lights, and electrical discharge sources. In some embodiments, the excitation component comprises a plurality of sources of electromagnetic radiation (e.g., a plurality of LEDs, OLEDs, flash bulbs, emissive species, and / or electrical discharge sources). In some embodiments, the source of electromagnetic radiation comprises paired light-emitting diodes. In some cases, two or more sources of electromagnetic radiation are configured to emit electromagnetic radiation in the same range of wavelengths. In some instances, each electromagnetic radiation source of the plurality of electromagnetic radiation sources is configured to emit electromagnetic radiation in the same range of wavelengths. In some cases, two or more sources of electromagnetic radiation are configured to emit electromagnetic radiation in different ranges of wavelengths. In some instances, each electromagnetic radiation source of the plurality of electromagnetic radiation sources is configured to emit electromagnetic radiation in different ranges of wavelengths.

[0096] #14629375vl In some embodiments, the sensor is configured to determine one or more characteristics of the plurality of cells based upon a signal from the impendence detector and / or the one or more photodetectors as a fluid (e.g., comprising the plurality of cells) flows through the fluidic channel. In some embodiments, the signal from the sensor is correlated with two or more characteristics selected from the group consisting of cell viability, cell concentration, and / or cell type.

[0097] The sensor may comprise any suitable number of optical components (e.g., one or more optical components, two or more optical components, three or more optical components) and / or any suitable number of sources of electromagnetic radiation (e.g., no sources of electromagnetic radiation, one or more sources of electromagnetic radiation, two or more sources of electromagnetic radiation, three or more sources of electromagnetic radiation).

[0098] The device may be assembled using any suitable methods. In an illustrative set of embodiments, and without wishing to be bound by such, an exemplary system comprises two impedance sensors and two pairs of optical sensors to collectively determine the cell viability, concentration, and cell type during therapeutic administration. In some embodiments, collected signals are further processed with customized machine learning models to confirm accurate detectability of known concentrations and viability, as well as predict unknown concentration and viability simultaneously.

[0099] In an exemplary set of embodiments, fabrication of this device involves two steps: (1) build multi-modal sensors on a thin film substrate and (2) integrate the multi-modal sensors in an “In-line” module that is compatible with commercial syringes and needles. For example, in some embodiments, laser ablation micro-machining is performed on a flexible tri-layer copper (18 micron) - polyimide (75 micron) - copper (18 micron) printed circuit board (fPCB) to form the impedance sensors, soldering pads for optoelectronic components and electrical connectors, and interconnection traces (see e.g., FIG. lD(a)-(f)). In some embodiments, a chemical vapor deposition (CVD) process then conformally encapsulates the device with ~13 microns of parylene. In some embodiments, a second laser ablation of the parylene exposes the electrodes of the impedance sensors as well as the soldering pads for optoelectronic components and electrical connectors (FIG. ID(c)). In some embodiments, an external polyimide (PI, 75 microns) shadow mask is micromachined and microscopically aligned to the fPCB to cover all the components except the rectangular electrodes of the impedance sensors (FIG. ID(d)). In some embodiments, an electron beam vapor (E-beam) deposition process then sequentially deposits titanium (Ti; 10 nm) and gold (Au; 200 nm) on the exposed electrodes (FIG. lD(e)-(g)). In some embodiments, a

[0100] #14629375vl hot-air soldering bonds two micro-inorganic light emitting diodes (micro-ILEDs) with 470 and 535 nm wavelengths and two micro-inorganic photodiodes (micro-IPDs) as the components for the optical sensors (FIG ID(h)). In some embodiments, a dip coated optical epoxy (Norland 61) followed by ultraviolet (UV) light curing seals and mechanically strengthens the soldering joints of micro-ILEDs and micro-IPDs. In some embodiments, the fPCB is then assembled into smart needle prototypes (see e.g., FIG. 4B).

[0101] In some embodiments, the system further comprises a lumen and a distal end comprising a tip geometry suitable for intravenous insertion into a subject (e.g., a needle). In some embodiments, the impedance detector and / or optical component embedded within the needle. The lumen of the needle may be any dimension known to those of skill in the art. For example, in some cases, the lumen dimension may be chosen to correspond to standard needle gauge dimension. Without wishing to be bound to any particular theory, those of skill in the art will know that needles are commonly sized by gauge (e.g., 10 gauge (G) needle, 11G, 12G, 13G, 14G, 15G, 16G, 17G, 18G, 19G, 20G, 21G, 22G, 22sG, 23G, 23sG, 24G, 25G, 25sG, 26G, 26sG, 27G, 28G, 29G, 30G, 31G, 32G, 33G, or a 34G needle) with each gauge having a standardized external diameter and length. For example, a standard 14G needle typically has an external diameter of 1.83 mm whereas a 18G needle has a 1.27 mm external diameter.

[0102] Any tip geometry known to the skilled artisan may be used. Non-limiting embodiments include a bevel tip needle stylet, bevel tip needle cannula, lancet point needle stylet, back bevel needle stylet, back bevel needle cannula, trocar tip needle stylet, Franseen tip needle cannula, or conical tip needle stylet. In some embodiments, the tip comprises a symmetric geometry; however, in certain embodiments the tip comprises an asymmetrical geometry.

[0103] In an exemplary set of embodiments, the system comprises a fluidic connector comprising an inlet, the fluidic connector sized and adapted to mechanically interface with a medical device, if present, a fluidic channel in fluidic communication with the inlet, a flexible sensor positioned proximate the fluidic channel and comprising one or more photodetectors and an impedance detector, the impedance detector comprising two or more electrodes, wherein the impedance detector is in electrical communication with the fluidic channel. In some embodiments, the medical device is a syringe.

[0104] In another exemplary set of embodiments, the system comprises a fluidic component configured for the flow of a plurality of cells, the component comprising an inlet, a fluidic channel in fluidic communication with the inlet, a sensor positioned proximate the fluidic

[0105] #14629375vl channel, wherein the sensor is configured to substantially simultaneously determine two or more of a cell concentration, a cell viability, a flow rate, and a cell type of the plurality of cells.

[0106] In some embodiments, the systems described herein may be useful for determining one or more characteristics of a plurality of cells. In some embodiments, a fluid comprising the plurality of cells is flowed through a fluidic channel such that the plurality of cells flows past a sensor, the sensor comprising one or more photodetectors and an impedance detector. In some embodiments, the method comprises measuring, via two or more electrodes of the impedance detector, an impedance of the fluid as it flows through the fluidic channel, thereby generating an impedance signal. In some embodiments, the fluidic channel is exposed to a source of electromagnetic radiation and the one or more photodetectors detects a change in the electromagnetic radiation, thereby generating an optical signal. In some embodiments, a characteristic of the plurality of cells is determined based upon the impedance signal and the optical signal. In some embodiments, two or more characteristics are determined substantially simultaneously based upon the impedance signal and the optical signal.

[0107] In some embodiments, the system is reusable. For example, in some embodiments, the reusable system has the capacity to integrate on an inner surface of a needle (or an adapter associated with the needle), the system is configured to measure one or more of cell viability, concentration, and flow speed during flow of a fluid through the needle.

[0108] Electronic components may be any suitable electronic component known to the skilled artisan. In some embodiments, the article may comprise an electrical system. In some embodiments, an electrical system may include two or more electronic components. For instance, in some embodiments, the electrical system comprises a power source (e.g., a battery as described above), an actuator such as an electrical actuation control system, a microcontroller, a PCB, a wireless component, a central processor, a power management system, a system wakeup controller, and / or one or more electronic sensors such as temperature sensors, a microphone, and / or humidity sensors. In some embodiments, the electronic component comprises a temperature sensor. In some embodiments, the electronic component comprises a microphone. In some embodiments, the electronic component comprises a pH sensor. In some embodiments, the electronic component comprises a carbon dioxide sensor. In some embodiments, the electronic component comprises an accelerometer. In some embodiments, the electronic component comprises an oxygen sensor. In some embodiments, the electronic component comprises a hydrogen sulfide sensor. In some embodiments, the electronic component comprises a pressure sensor. In some embodiments, where the electronic component comprises a sensor (e.g., a

[0109] #14629375vl temperature sensor, a pressure sensor, an oxygen sensor, etc.), the sensor is configured to operate within a liquid, e.g., at physiological conditions.

[0110] In some embodiments, the one or more electronic components are in electrical communication with another component of the article or system. Any electronic component circuitry may be implemented by any suitable type of analog and / or digital circuitry. For example, the electronic component circuitry may be implemented using hardware or a combination of hardware and software. When implemented using software, suitable software code can be executed on any suitable processor (e.g., a microprocessor) or collection of processors. The one or more electronic components can be implemented in numerous ways, such as with dedicated hardware, or with general purpose hardware (e.g., one or more processors) that is programmed using microcode or software to perform the functions recited above.

[0111] In this respect, it should be appreciated that one implementation of the embodiments described herein comprises at least one computer-readable storage medium (e.g., RAM, ROM, EEPROM, flash memory or other memory technology, or other tangible, non-transitory computer-readable storage medium) encoded with a computer program (i.e., a plurality of executable instructions) that, when executed on one or more processors, performs the abovediscussed functions of one or more embodiments. In addition, it should be appreciated that the reference to a computer program which, when executed, performs any of the above-discussed functions, is not limited to an application program running on a host computer. Rather, the terms computer program and software are used herein in a generic sense to reference any type of computer code (e.g., application software, firmware, microcode, or any other form of computer instruction) that can be employed to program one or more processors to implement aspects of the techniques discussed herein.

[0112] In some embodiments, the one or more electronic components are connected to the article using one or more techniques known to those skilled in the art. For example, in some embodiments the one or more electronic components are mechanically press fit into the system. In other embodiments, the one or more electronic components are connected to the system using a bonding method. In some embodiments the bonding method is any suitable bonding method known to those of skill in the art. For example, in some embodiments the bonding method comprises chip bonding. In other embodiments, the bonding method comprises wire-tacking adhesives. In some embodiments, the bonding method comprises potting an encapsulation. Without wishing to be bound by any particular theory, potting is a method known in the art that comprises the filling small spaces or services with the material that will protect components from

[0113] #14629375vl physical and environmental damage. Typical resins used for potting include epoxies and silicones, some of which may be UV-curing formulations. Other resins are also possible. In some embodiments, encapsulation comprises casting and / or molding electronic component using similar resins as described above. In some embodiments, the bonding method uses an adhesive. In some embodiments, the adhesive comprises an electrically conductive adhesives, a thermally conductive adhesives, or an UV-curing adhesives. Other adhesives are also contemplated in some embodiments, for example, cyanoacrylates, silicone resins, and polyimides.

[0114] The following examples are intended to illustrate certain embodiments of the present invention, but do not exemplify the full scope of the invention.

[0115] EXAMPLES

[0116] A smart cell analysis needle system (SCANS) capable of determining the viability, concentration, and type of cells at the point of administration is described. The SCANS incorporates impedance and optical sensors that can analyze a cell suspension in real time through the application of flexible electronics, allowing its convenient deployment in series with injectable system. Application of customized machine learning models in SCANS effectively improved the range and resolution of cell analysis by realizing accurate prediction of cell characteristics. The SCANS approach can be readily applicable during cell therapy administration as well as permitting enhanced quality control in the manufacturing of these sensitive but potent cell therapeutics.

[0117] FIG. 1C shows a schematic of a smart cell analysis needle system (SCANS), according to one set of embodiments. Further description of FIG. 1C is as follows. Accurate evaluation of cell therapy generally uses rapid real-time detection of cell characteristics including viability, concentration, and type during therapeutic administration. Traditional methods cannot satisfy these requirements due to their spatiotemporal restrictions as they require operations of bulky equipment and microscopes to analyze prepared cell samples. Our proposed smart cell analysis needle system (SCANS) equips multimode sensing system that contains impedance sensors as well as paired light emitting diodes (470 and 525 nm) and photodiodes to detect cell characteristics. A flexible printed circuit board (fPCB) substrate holds all functional components and rotates around a customized transparent shrink wrap which is deployed between a cut commercial Luer lock to form the in-line sensing module. The sensing module connects the syringes and injection needles and is compatible with all commercial sizes. During the injection delivery of cells to multiple locations of the body (brain, heart, liver, pancreas, etc.), a

[0118] #14629375vl customized machine learning model incorporating artificial intelligence outputs the viability, concentration, type of the cells based on the signals collected from the sensing module. In addition, this system can also provide rapid monitoring of cell status during preparation, packaging, and transportation of cells.

[0119] This example describes the development of a smart cell analysis needle system (SCANS) through a platform that can provide real-time, rapid, and accurate assessment of cells at the point of transplantation without the need to prepare cell samples and dissociate the cell clusters. As demonstrated in FIG.l, the SCANS contains multi-mode sensing module that characterizes the viability and concentration as well as the type of the cells, a machine learning (ML) data analysis model, and a modular assembly setup for deployment in-line between commercial needles and syringes with arbitrary sizes. The sensing electronics contains two impedance sensors and two optical sensors with each made from a paired microscale light emitting diodes ( .LED) and photodiodes (|iPD). The sensors are affixed at the surface of a thin flexible printed circuit board (fPCB; - 140 pm thick) which is curled around a transparent shrink wrap and assembled with commercial Luer lock syringe and needle adapters. The impedance and optical sensors exploit cell type and status-dependent electrical and optical properties to distinguish their type, viability, and concentration during the injection deliveries towards various body locations. Customized machine learning (ML) models derived from thorough statistical analysis improve the accuracy of cell characterization by combining the impedance and optical measurements through an attention mechanism with temporal- spatial featuring. Meanwhile, the artificial intelligence (Al) introduced by the calibration data-trained ML models allow statistically outstanding predictions, defined by around or below 5% mean absolute percentage error (MAPE) of cell viability and 30% concentration that are both not included in the calibration dataset. In addition, the SCANS can also serve as a convenient and rapid monitor of cell characteristics during preparation, packaging, and transportation of cells. SCANS was tested with various types of cells in both single cell and cluster form, demonstrating its wide adaptability in different clinical trials. The entire technology may require standard procedures that widely adopted in fPCB industry, which allows broad dissemination of this technology in daily clinics in near future.

[0120] Impedance characterization of cell viability and concentration in static scenario

[0121] The electrical properties of cells and medium as well as their relative volumes in suspension collectively determine the electrical properties of the cell suspension. Meanwhile, the electrical properties of individual cells are determined by their types and viabilities. Thus, when

[0122] #14629375vl applying an electrical field in cell suspension, the measured impedance reflects the viability, concentration, and type of cells via the overall resistive and capacitive behaviors of the cell suspension (FIG. 2A). Here, a thin planar impedance sensor was used as a proof-of-concept for prototyping the SCANS. The thin film impedance sensor (FIG. 2B) contains Cr (10 nm) / Au (200 nm) electrodes lithographically defined in rectangular shapes and toothing arrangement embedded in symmetrical polyimide (PI) films (-2 m). The top PI layer has patterned windows that expose the electrodes and contact pads (FIGS. 2B and 2C). An underneath soft silicone substrate (-200 pm) holds the thin film sensor and prevent it from curling and wrinkling (FIG. 2B). An impedance analyzer connects to the thin film sensor and collects the impedance of different 10 pL cell suspensions statically with known viabilities (from 2.3% to 94.3%) and concentrations (from 10 k to 10 M cells / mL). In this experiment, human embryonic stem cells (hESC) were used in single cell form. More details on the fabrications and experimental procedures can be found in the FIG. 6.

[0123] Three impedance sensor designs with lateral dimensions of 0.5, 1.0, and 1.35 mm were tested (FIG. 7-16). The representative design had a planar size of 1.35 x 1.35 mm2with 7 pairs of rectangular electrodes (1.35 x 0.05 mm2) with 0.05 mm interspacing (FIG. 2C). FIG. 4D shows the resistive (real part of the impedance; top), the capacitive (imaginary part of the impedance; middle), and the absolute values (bottom) of the impedance of hESC single cells with different viabilities (86.0%, 77.1%, 54.0%, 43.1%, 13.1%, and 2.3%) at constant 10 M cells / mL concentration scanned between 100 Hz to 1 MHz at a logarithmic scale. Results show increase in both resistive and capacitive behavior in solution with respect to increased cell viabilities across most of the selected frequencies, respectively. Meanwhile, the absolute values of the impedance increase with respect to increased cell viabilities across most of the selected frequencies.

[0124] An analysis of covariance (ANCOVA) was applied to evaluate the performance of the thin film impedance sensor at different frequency range to seek the optimal frequency range to differentiate cell viability and concentration with minimum scans, aiming to reduce the electrical damage to the cells. In consistence with mathematical theory, the terminology “statistical effect” was used to refer to variables of interest for analysis. Specifically, it was considered how each statistical effect including frequency, viability, concentration, and their interactions, contributes to the variance (R2score) of the resistive part, capacitive part, and the absolute values of the collected impedance. Further, types of sum of squares were tested to evaluate the significance of each effect and their interactions to the explanation of variance of detected impedance. Here, Type III partial sum of squares was used, which was calculated to consider the partial

[0125] #14629375vl significance of each statistical effect, inclusively appraising every other effect in the ANCOVA model. FIG. 4E shows that the impedance sensor was generally ideal for detecting the differentiation of cell viability, as frequency, viability and their interactions were highly significant (<0.05) at nearly all frequency range to differentiate both capacitive and the absolute values of the impedance. However, the impedance sensor generally helped with differentiating variance of impedance with respect to the concentration at a low scanning frequency range from 100 - 2000 Hz (FIG. 4F). Further, ANCOVA model indicated that the impedance sensor was able to increasingly differentiate capacitive part as well as the absolute values of the impedance in terms of cell viability and concentration when cumulatively enlarging the frequency range up to 5 kHz, while increasing the frequency scanning range may not help differentiate the resistive behaviors in response to change of cell viability and concentration (FIGS. 4G & H).

[0126] Optical characterization of cell viability and concentration in static scenario

[0127] Although impedance measurements provide accurate assessments of cell viability and concentration, the cross-interferences between these two factors may cause inaccuracies during applications when both factors are unknown. A separate sensing mechanism is important to achieve improved accuracy on cell characterization for the SCANS. The optical properties of individual cells are collectively determined by their types and viability. Thus, optical measurement that exploits the absorption and scattering of photons when interacting with cells in suspension becomes a promising solution. A microplate reader generating an absorption spectrum (400-1000 nm) from straight incident light beams that pass through the cell suspension (FIG. 3 A) provides initial insights into optical detection of cell concentrations and viabilities. FIG. 3B shows the absorption spectrum of HEK293 cells in single cell form with -100% constant viability at different concentrations ranging from 100k to 8 M cells / mL. FIG. 3C shows the absorption spectrum of HEK293 single cells with 1 M cells / mL concentration at viabilities of 0%, 20%, 40%, 60%, 80%, and -100%.

[0128] ANCOVA was once again performed to characterize the performance of the optical sensing system and determine the optimal wavelength range in differentiating cell viability and concentration. Here, the ANCOVA model considered how cell viability, concentration and photonic wavelength contributed to the variance of detected current change from the photodiodes in the microplate reader. FIGS. 3D and 3G describe how much variance of current change with respect to different cell concentrations and viabilities can be explained under different wavelength scanning range, respectively. Results show that blue and green light (wavelength <

[0129] #14629375vl 600 nm) is better than red and infrared light (wavelength > 600 nm) to differentiate current change in terms of different cell concentrations and viabilities. Further, both Type I sequential sum of squares and Type III partial sum of squares confirmed that that wavelength, cell viability and concentration are highly significantly (< 0.05) in differentiating current change of the sensor optical sensing system at the green and blue light range, but less significant in the red and infrared light range (FIGS. 3E, 3F, 3H and 31).

[0130] Characterization of cell viability, concentration, and type in SCANS during dynamic delivery

[0131] Advancement in fabrication technologies in fPCB industry may allow the transfer of the complex sensing systems adopted in bulky equipment for cell characterizations to a miniaturized, thin, and flexible device in simplified architectures. A device with impedance and optical sensors on a thin Cu (18 pm)-PI (75 pm)-Cu (18 pm) tri-layer fPCB was built to form the SCANS that can characterize the viability, concentration, and type of cells during dynamic delivery (FIG. 4A). Laser ablation on the top Cu side of the fPCB form two impedance sensors that are identical to those demonstrated in FIG. 2C. Electrodes (2.16 x 0.1 mm2, 0.06 mm interspacing) of the impedance sensors are covered with Ti (20 nm) / Au (200 nm) coatings on their surfaces for biocompatible purposes and isolated with parylene (~13 pm) on their side walls. Two pairs of microscale light emitting diodes (pLED; 470 and 525 nm wavelengths; 1.6 x 0.8 x 0.95 mm3) and photodiodes (pPDs;l x 1 x 0.65 mm3) located in alignment form the optical sensing system that mimic the set up in the microplate reader. An adhesive tape attaches the fPCB with impedance sensors and optoelectronic components exposed (FIG. 4B). In this way, four independent optical detection modes, marked by the photocurrent signals from either direct or scattered light at either 470 or 525 nm after penetrating through the cell suspension, were established (refer FIG. 18 for schematic illustration). The four independent detection modes collectively determine the optical cell characterizations. A commercial Luer lock that cut in half sandwiched the curled fPCB that encapsulated with two layers of heat shrink wrap and formed the sensing module of the SCANS. In this way, the sensing module was compatible with commercial interchangeable needles and syringes with different sizes (FIG. 4B). A 12-pin flexible flat cable (FFC) connected between two FFC connectors that located on the sensing module and a customized breakout board completes the SCANS (FIG. 20 A). During operation, a personal computer (PC) controls an impedance analyzer and a microcontroller (pC) that both connected to the breakout board to transmit the operation commands and collect corresponding

[0132] #14629375vl signals. Two transimpedance amplifiers convert the current signals from pPDs to voltages and filter the noises beyond 10 kHz. This connection mode minimized the electrical disturbances from the ambient environment and wiring during operation. FIGS. 20C and 20D shows the impedance data of distilled water collected at 1, 5, 10, 100, and 1000 kHz after random motions, positions, and repetitive connections. The variations fall in the range between 0.5%~ 1.5%, matching the data variations from static impedance tests, indicating consistently accurate detection of cell characteristics regardless of external factors such as wiring and positioning. FIG. 17-20 shows details on the fabrication, assembly, and operation of the SCANS. In addition, FIG. 21 shows an option of integrating the impedance sensors into a 10-gauge commercial needle for future miniaturization of the SCANS.

[0133] The detection of the differentiation of hESC was tested in both single cell and cluster forms as well as BC09 and cardiomyocyte clusters at different viabilities and concentrations during dynamic delivery through our SCANS. Mixture of cells at -100% viability and 0% viability yield single cell suspensions at different viabilities. Time-controlled immersion of cell clusters in dimethyl sulfoxide (DMSO) solution mimicked the natural apoptosis of cells in cluster form and yield cell cluster suspension at different viabilities (FIG. 22-23). Dataset from impedance sensors was collected. Dataset from optical sensors is summarized in FIG. 24-29.

[0134] Analysis in this section focuses on the data structure of the dynamic measurements to elucidate the design strategies for building up ML models that enhance the detecting range and accuracy of the system. Compared to the static condition, data collected by the SCANS during dynamic delivery exhibited higher nonlinearity, which indicates that ANCOVA was not appropriate anymore. Thus, a spearman correlation analysis was adopted to evaluate how each statistical effect was related to each other in the impedance and optical measurements. Results from the hESC single cells serve as a representative. For optical measurement, cell viability and concentration were more directly correlated with the detected current change in the optical sensor, while the voltage change and detection modes were indirectly correlated to cell viability and concentration through detected current (FIGS. 4C-4E). For the impedance sensor, cell viability and concentration were equivalently correlated to every single effect in terms of the impedance measurement, including the resistive part, capacitive part, absolute value, and phase angle of the impedance (FIGS. 4F-4H). In addition, classification of dynamic detection data with respect to cell viability and concentration demonstrated that both optical and impedance sensors perform well in differentiating cells under different viability and concentration, whereas different classes of cell concentration and viability were vaguely overlapped with each other. Furthermore,

[0135] #14629375vl sequential scanning of both optical and impedance sensor depicted a temporal-spatial difference in their measurements. Such difference will generally be undermined through classical classification and similarity analysis. Hence, a dynamic time warping (DTW) algorithm was adopted, which is well-known for speech recognition, to consider the temporal-spatial similarity in recognizing various cell types through a dynamic detection set-up. The results show that both optical and impedance sensors can well identify different cell types between each other (similarity < 0.6; FIG. 33), and a combination of both detection approach can lead to even better performance in most cases (FIGS. 41 and 4J).

[0136] Machine learning-assisted SCANS for augmented detection range and accuracy

[0137] To compensate for the loss in detection capacity of the SCANS due to miniaturization, ML was adopted to improve the detection response time, accuracy, and generalization ability to detect unseen cell viability and concentration. The hESC single cells, shown in FIG. 4, served as a representative to exemplify the capability of ML-assisted detection for the SCANS. As demonstrated in FIG. 4, the data availability from the SCANS may be limited, especially for those collected by optical sensors. Therefore, the ML model may cater to the small size of training dataset, and the large difference in target variables of cell viability and concentration. Here, regression models were built that predict the cell concentration and viability based on the measurement of the impedance and optical sensors. All dataset of cell viability were held off at 90% and concentration at 20 M cells / mL for testing, and used the rest data for training to demonstrate the capacity of the ML models in interpolating the highly-dimensional model between the target variables (cell viability and concentration) and predictor variables (resistive and capacitive part of the impedance, scanning frequency, wavelength, detection mode of optical sensing and experimental repetitions). Further, feature generation was conducted by including the absolute value and phase angle of the detected impedance to enhance the fitting performance of the ML model.

[0138] A single regression model of the two target variables was built for the optical and impedance sensor, respectively, to evaluate the measurements of each type of sensors for prediction. Tree-based random forest models (FIG. 5 A) were used, considering its benefits in regression and generalization on small datasets. To help ML model understand the same impedance measurements shared between the target variables, a special bouncing system to pass information in between regression of cell viability and concentration was built to guarantee that the tree model made a global decision in splitting the nodes. Considering the magnitude

[0139] #14629375vl difference in the target variables, especially in cell concentration, a mean absolute percentage error (MAPE) was used, which depicts the absolute percentage ratio of the prediction in terms of the true value, to evaluate the performance of the ML models. Regression results show that the prediction MAPEs for the optical sensors are smaller or around 10% for all training datasets in predicting both cell viability and concentration (FIG. 5C). For the test dataset, both optical and impedance models can still maintain an accuracy of 10% MAPE for most of the viability prediction, while the MAPE of predicting concentration may increase to 50% (FIG. 5C & 5D). The prediction of training datasets performed even better on the impedance sensor-only model, of which MAPE is smaller than 1% (FIG. 5D). This indicates that the ML-assisted SCANS can accurately predict any cell viability and concentration that it knows through detection training. Further, a model-based on optical sensor may perform better than the impedance sensor in predicting concentration, and vice versa for viability, which is consistent with our findings in characterizing the two types of sensors in static scenarios demonstrated in FIG. 2 and 3. In addition, MAPEs between train and test data is generally large, indicating that the generalization of single sensor-based model can be further improved by combining them together.

[0140] To combine the detection merits of both optical and impedance models, an attention ML model was built to jointly predict based on the measurements of both sensors. To exhaust the temporal- spatial features of sequential frequency and wavelength-based scanning of the impedance and optical sensors, a time-series attention model was introduced, where a time-series gate recurrent units (GRU) model was separately built for regressing the optical and impedance model separately, and a multilayer perception (MLP) transformer networks was then built to jointly predicting cell viability and concentration based on both optical and impedance responses (FIG. 5B). Results show that the generalization of the attention model has been significantly improved, compared to the single sensor-based models, where the MAPEs of predicting test datasets were as low as 5% for viability and 20-30% for concentration (FIG. 5E). Note that there is one test prediction outlier at viability of 90% and concentration of 2 M cell / mL. The presence of this outlier is due to the intrinsic sparsity of the concentration data, where values are of multiplication relation to each other. Such sparsity involves difference in magnitude of order which in theory leads to low performance of the model. However, the actual prediction error of this data point is about 0.9 M cell / mL, which can be acceptable. In addition, such outlier can be significantly improved when more data is sampled in clinical trials. Further, most of the prediction MAPE has decreased by two to three times in the test datasets, compared with the single sensor-based models. These results indicated that ML-assisted SCANS possesses ideal

[0141] #14629375vl generalization capacity in predicting combination of cell viability and concentration that it has never seen before. More than that, such predication accuracy and capacity can be further improved by introducing more scanning frequencies and wavelengths in the SCANS (FIG. 5F- 5G).

[0142] Discussion

[0143] Accurate assessment of the cell type, viability, and concentration of cell suspensions at the site of administration in real-time stands to transform our capacity to inform the efficacy of cell therapy-based therapeutic interventions. Moreover, a facile portable system like SCANS enables rapid evaluation of the cell product during the manufacturing process which helps the quality control process associated with the complex and sensitive therapeutic class. Other methods on characterization of cells rely on complex sample preparations and bulky equipment which cannot satisfy neither spatial nor temporal requirement for real-time rapid monitoring. A multimode in-line module that deployed between needles and syringes with important cell characterization functions may be a solution. It was demonstrated through SCANS that the cells’ responses can be applied to electrons and photons through sensing systems on a fPCB that can be conveniently curled around a customized commercial Luer lock. A comprehensive data set combined with statistical analysis of variance confirmed the accurate detection of the type, viability, and concentration of cells in both single cell and cluster forms during dynamic delivery with the impedance and optical sensors from the SCANS. Meanwhile, the introduction of artificial intelligence (Al) in the form of customized machine learning (ML) models allow the SCANS to predict the viability and concentration simultaneously and accurately from data excluded from training datasets. This ML-based prediction algorithm remarkably improves the resolution of the SCANS and reduces the size of the dataset required for calibration, which collectively facilitate the translation of this technology in daily clinics. The results from the ML- based prediction suggest improved accuracy on prediction with respect to the increased diversities of detection modes of cell characteristics. This phenomenon inspires the future development of the SCANS. By incorporating additional characterization mechanisms such as fluorescence and biochemical sensing as well as introducing more wavelengths into optical sensing, the SCANS is expected to achieve greater accuracy with respect to predictions on cell characteristics. Additionally, advancement in micro / nanofabrication and industrial manufacturing stands to support even further reduction in size of the SCANS.

[0144] #14629375vl Method

[0145] Thin-film flexible impedance sensor fabrication

[0146] This process utilized standard cleanroom facilities. The fabrication began by spin coating polyimide (PI; 2000 rpm, 30 s, ~ 2 pm thickness) on a 4-inch thermal oxide wafer followed by curing process performed at a vacuum oven set at 200 °C. Electron beam vapor (E-Beam) deposition formed thin Cr (10 nm) / Au (200 nm) metallic layers sequentially on PI. Photolithographical patterning of the metallic layers began by sequential spin coating of hexamethyldisilazane (HMDS) and SI 805 photoresist followed by 1 -minute soft bake on a hotplate set at 110 °C. Ultraviolet (UV) light exposure using a mask aligner (40mJ dose) followed by developer rinse (CD-26) formed the pattern of the impedance sensors on the photoresist. Sequential etching using Au and Cr etchant printed the pattern from the photoresist to the underneath metal layers. After the removal of the photoresist and HMDS using acetone rinse and reactive-ion etch (RIE), a second PI layer thickness of ~ 2 pm was spin coated and cured on top of the patterned metal layers following the same process as demonstrated in the first PI layer coating. Then, a second photolithographical patterning began by sequential spin coating and soft baking of HMDS, LOR3A (5-minute hotplate bake at 110 °C), and S1805 photoresist (1- minute hotplate bake at 110 °C) followed by the same UV patterning and developing procedures to form a photoresist layer that covered the electrodes and contact pads of the impedance sensors. Acetone rinse after thermal evaporation of Ni (60 nm) on the surface formed a Ni hard mask with electrodes and contact pads of the impedance sensors exposed. RIE process removed the PI on the electrodes and contact pads. Finally, a water-soluble tape transferred all devices from the wafer substrate after removing Ni hard mask with Ni etchant. Please refer to FIG. 6 for graphical demonstrations. fPCB device fabrication

[0147] Laser ablation performed on the top surface of a tri-layer copper (18 pm) - polyimide (PI; 75 pm) - copper (18 pm) flexible printed circuit board (fPCB) formed the device substrate that contained impedance sensors, soldering pads for optoelectronic components and a 12-pin flat flexible cable (FFC) connector, and interconnection traces. The back side copper served as a metallic shield to prevent electrical disturbance during the measurement. A chemical vapor deposition (CVD) process then conformally encapsulated the device with ~13 pm of parylene. A second laser ablation of the parylene coating exposed the electrodes of the impedance sensors as well as the soldering pads for optoelectronic components and the FFC connector. A laser ablated

[0148] #14629375vl PI (75 pm) shadow mask covered the fPCB with only rectangular-shaped electrodes of the impedance sensors exposed. Electron beam vapor (E-Beam) deposition process sequentially deposited Ti (20 nm) and Au (200 nm) on the exposed electrodes. Next, hot-air soldering bonded two microscale light emitting diodes (pLEDs) with 470 nm and 525 nm wavelengths and two microscale photodiodes (pPDs) as the components for the optical sensors. Dip coated optical epoxy cured by ultraviolet (UV) light sealed and mechanically strengthened the soldering joints of pLEDs and pPDs.

[0149] Device assembly

[0150] Assembly of the SCANS began by heating a fluorinated ethylene propylene (FEP) shrink wrap (0.385 inch, 2:1 shrink factor) onto a 6.61 mm assembly rod using a heat gun to create a tight-fitting transparent inner wall for the sensing module of the SCANS. Hand pressing flattened this shrink wrap after removal from the assembly rod. A 4 mm hole puncher cut two half-circular holes on the edges of the flattened wrap that later aligned with the impedance sensors on the fPCB for direct access to the solution medium that under investigation. The shrink wrap was then placed back on to the assembly rod. To note, the optical components can operate through the transparent inner wall. Thus, they may not need direct contact with the solution. A PI tape (Kapton) covered the backside of the fPCB to electrically insulate the backside copper of the fPCB from the environment.

[0151] Laser ablated very-high-bond (VHB) foam mounting tape (0.045 in thick) with designed openings for the impedance and optical sensors adhesively covered the front surface (component side) of the fPCB. The foam tape served as a leak proofing medium that prevented solution from passing through the impedance sensor holes in the inner shrink wrap. The fPCB with the foam tape was then wrapped around the assembly rod with impedance sensors aligned with the holes. After wrapping, two 3 mm pieces of wrapped foam tape on both edges of the rolled fPCB secured the location of the device from lateral movement. A second shrink wrap (High-Strength Heat-Shrink Polyvinylidene Fluoride Tubing, 0.5 inv, 2:1 shrink factor) sealed the entire device into a compact form factor after slowing (5 s) followed by long cooling cycles (40 s). This heat shrinking process of the outer shrink wrap prevented the melting of the inside foam tape. A cut window into the outer shrink wrap exposed the edge of the fPCB. The assembled sensing module was then pulled off the assembly rod and sandwiched between a commercial luer lock that cut in half from the middle. UV curable epoxy then glued the luer lock with the sensing module to fix the relative position. Finally, a soldered 12-pin FCC connecter on the exposed edge of the fPCB

[0152] #14629375vl connected all sensors with external hardware. By then, this sensing module can connect to commercial syringes and any gauge needles to form SCANS for cell delivery.

[0153] Cell culture and characterization

[0154] INS-1 832 / 13 rat insulinoma cell line (Cat. SCC207) were used. Cells were grown and expanded in RPMI-1640 medium supplemented with 2 mM L-Glutamine, 1 mM sodium pyruvate, 10 mM HEPES, 0.05 mM / ? -mercaptoethanol and 10% FBS. Additionally, 100 U / mL penicillin and 100 pg / mL streptomycin at 37 °C in a humidified incubator containing 5% CO2. For the cluster formation, INS-1 cells were seeded at a density of 100 x 106in 500 mL polycarbonate Erlenmeyer flask with total medium volume 50 mL. The flask was placed on the orbital shaker speed 70 rpm, cultured for 48 to 72 hours. Human embryonic stem cell (WA01, HUES8) lines were maintained in mTeSR+. HL-1, HEK293T, Neuro2A, bEnd.3, HEPG2, Caco- 2 cell lines were maintained in claycomb, DMEM, EMEM medium. hESCs and cardiomyocytes.

[0155] For the single cell viability control, cells were immersed in DMSO solution for 30 minutes to make 0% viability. Mix quantified ratio of 100% and 0% samples to make different viability. For the INS-1 cluster viability control, clusters were immersed in 50% DMSO solution for 1, 5, 10 and 15 minutes. For the hESC cluster viability control, clusters were immersed in 30% DMSO solution for 2, 5 and 10 minutes, 40% DMSO solution for 5 minutes. For the cardiomyocyte viability control, clusters were immersed in 30% and 40% DMSO solution for 2 minutes.

[0156] Characterization of the viability and concentration of single cells at static scenario

[0157] Human embryonic stem cells (hESC) in single cell form were used for impedance detection of the viability and concentration in static condition. A customized printed circuit board (PCB) electrically connected the thin film sensors with an impedance analyzer (Sciospec). All thin film sensors were cleaned with distilled water (Dl)-wetted and dry cleaning wipes (KimTech) sequentially before each measurement. A variable volume single channel pipette selected at 10 pL collected the cell suspension with controlled viability and concentration, and then dropped on the surface of the thin film sensor. The impedance analyzer immediately generated a constant 500 mV across the positive and negative electrodes of the thin film sensors and collected the corresponding impedance for three repetitions at 1024 individual frequencies between 100 Hz and 1 MHz selected in logarithmic scale. Each experiment was repeated for three times.

[0158] #14629375vl HEK293 single cells were used for optical detection of the viability and concentration at static scenario. The culture medium was replaced by phosphate buffered saline (PBS) right before the experiment. A micro plate reader outputted the light absorptions of individual 100 pL cell suspension with controlled viability and concentration from 200 to 1000 nm wavelengths. Each experiment was repeated for three times. Data corresponding from 400 to 1000 nm wavelengths were selected for analysis.

[0159] Characterization of the viability, concentration, and type of single cells and clusters during dynamic flow

[0160] A microfluidic pump carried a syringe equipped with the SCANS and loaded with cell suspension at controlled viability, concentration, and type. A 12-pin FFC electrically connected the SCANS with a customized printed circuit board (PCB) via corresponding FFC connectors. The PCB served as a breakout board for the connections of two impedance sensors to external impedance analyzer, two pLEDs to corresponding drivers, and two pPDs to the inputs of two external transimpedance amplifiers. A microcontroller running custom Fab VIEW software digitally switched the blue and green pLED drivers to alternatively light up. The current amplifier converted photodiode current to voltage with a gain of 5,10, or 20 pA / V, which was read by the microcontroller analog to digital converter and recorded at 20 ms period onboard. A PC connected to both the microcontroller and the impedance analyzer via USB and controlled their operations via corresponding software. During operation, the impedance analyzer immediately generated a constant 500 mV across the positive and negative electrodes of the impedance sensors and collected the corresponding impedance for 10 repetitions at 200 individual frequencies between 1 kHz and 1 MHz selected in logarithmic scale. Each experiment was repeated for three times. The pPDs consistently collected photocurrent signals with pLEDs interchangeably blinked for 200 ms.

[0161] Statistical analysis

[0162] Analysis of covariance (ANCOVA) package in commercial software SAS Studio was used to evaluate contribution of each statistical effect to the covariance. In the ANCOVA model, cell viability and concentration were treated as categorical independent variables of different levels, while frequency and wavelength were treated as continuous independent variables. Furthermore, impedance and current change were treated as continuous dependent variables. To understand the optimal frequency for differentiating cell concentrations and viability, as well as

[0163] #14629375vl satisfying the linear regression assumption of ANCOVA model, frequency range of interest was disseminated as 100 - 1kHz, Ik - 2k Hz, 2k - 3k Hz, 3k - 4k Hz, 4k - 5k Hz, 5k - 20k Hz, 20k - 1MHz. Similarly, wavelength range of interest was disseminated as 400 - 500 nm, 500 - 600 nm, 600 - 700 nm, 700 - 800 nm, 800 - 900 nm and 900 - 1000 nm. For Type I and Type III sum of square that test the significance of each main effect and their interactions to the ANCOVA model, significance level a = 0.05 was used. Spearman correlation was implemented through open- source language Python 3.11.4, package pandas. Dynamic time warping (DTW) for differentiating of cell types through temporal-spatial recognition was performed through open- source language Python 3.11.4, package dtaidistance.

[0164] Machine Learning

[0165] Random forest model was built through open-source language Python 3.11.4, package scikit-learn. To seek the optimal performance of the model, 5-fold randomized search cross validation was adopted to tune the hyperparameters of the tree model. Specifically, the optimal number of trees in the model was tuned in a range from 10 to 1000, the max number of levels of the trees was tuned in a range from 10 to 110, the minimum number of samples required to split a node was chosen from the groups from 2 to 10, the minimum number of samples required at each leaf node was chosen from groups of 1 to 4. In addition, the number of iterations for hyperparameter optimization was tuned in range of 10 to 100 to avoid overfitting.

[0166] Attention model was built through open-source language Python and Py Torch including the gated recurrent unit (GRU) and the multilayer perceptron (MLP). Since the viabilities and concentrations are all positive and sparsely distributed, a log transform was applied to both targets and MSE loss was subsequently used. The subset of the data consisting of either 90% viability or 20 M cells / mL were held out for the test set while the rest of the data was randomly split 80%-20% for training and validation respectively.

[0167] The bi-directional GRU was chosen for the impedance and optical encoders with an MLP prediction head. Each encoder had identical architecture and were trained to convergence on their respective datasets with a patience window of 10. Then, for each encoder, embeddings for each entry of their datasets were collected as the hidden state of the last GRU cell.

[0168] To train the MLP for alignment of the embeddings, embeddings referring to the same viability and concentration were paired. Embeddings corresponding to either 90% viability or 20 M cells / mL were once again held out for the test set. The train- validation split was also randomly chosen at 80% and 20% respectively. For the GRU encoders, hyperparameter tuning involved the

[0169] #14629375vl number of layers: [1, 2, 4, 8], hidden size [16, 32, 64, 128], and dropout [0.25, 0.5, 0.75]. The MLP consisted of multiple FC layers with a ReLU activation between them. Hyperparameter tuning for the MLP involved the number of layers [1, 2, 4, 8] and hidden layer dimension [64, 128, 256],

[0170] Table 1. Gated recurrent unit and multilayer perceptron hyperparameters and respective values.

[0171] #14629375vl

Claims

CLAIMSWhat is claimed is:

1. An in-line flow system component, the in-line flow component comprising: a fluidic channel configured to receive a fluid comprising a plurality of cells, a sensor associated with the fluidic channel, the sensor comprising: an impedance detector comprising two or more electrodes in electrical communication with the fluidic channel; and an optical component proximate the fluidic channel, wherein the optical component comprises one or more photodetectors, wherein the sensor is configured to determine one or more characteristics of the plurality of cells based upon a signal from the impendence detector and / or the one or more photodetectors as the fluid flows through the fluidic channel.

2. An in-line flow system component as in claim 1, wherein the characteristic is selected from the group consisting of cell viability, cell concentration, and / or cell type.

3. An in-line flow system component as in any preceding claim, wherein the optical component comprises a source of electromagnetic radiation.

4. A fluidic component, the component comprising: a fluidic connector comprising an inlet, the fluidic connector sized and adapted to mechanically interface with a medical device, if present; a fluidic channel in fluidic communication with the inlet; a flexible sensor positioned proximate the fluidic channel and comprising one or more photodetectors and an impedance detector, the impedance detector comprising two or more electrodes, wherein the impedance detector is in electrical communication with the fluidic channel.

5. A fluidic component as in claim 4, wherein the medical device is a syringe.

6. A method for determining a characteristic of a plurality of cells, the method comprising:#14629375vlflowing a fluid comprising the plurality of cells through a fluidic channel such that the plurality of cells flows past a sensor, the sensor comprising one or more photodetectors and an impedance detector; measuring, via two or more electrodes of the impedance detector, an impedance of the fluid as it flows through the fluidic channel, thereby generating an impedance signal; exposing the fluidic channel to a source of electromagnetic radiation and detecting, via the one or more photodetectors, a change in the electromagnetic radiation, thereby generating an optical signal; and determining the characteristic of the plurality of cells based upon the impedance signal and the optical signal.

7. A reusable system with capacity to integrate on an inner surface of a needle, the system configured to measure one or more of cell viability, concentration, and flow speed during flow of a fluid through the needle.

8. A fluidic component configured for the flow of a plurality of cells, the component comprising: an inlet; a fluidic channel in fluidic communication with the inlet; a sensor positioned proximate the fluidic channel, wherein the sensor is configured to substantially simultaneously determine two or more of a cell concentration, a cell viability, and a cell type of the plurality of cells.

9. A fluidic component, method, or reusable system as in any preceding claim, wherein the impedance detector comprises one or more gold electrodes.

10. A fluidic component, method, or reusable system as in any preceding claim, wherein the gold electrode is fabricated lithographically.

11. A fluidic component, method, or reusable system as in any preceding claim, wherein the impedance sensor has a thickness of less than or equal 200 nm.#14629375vl12. A fluidic component, method, or reusable system as in any preceding claim, wherein the impedance detector is embedded in polyimide.

13. A fluidic component, method, or reusable system as in any preceding claim, wherein the impedance detector is configured to be operated at a frequency of greater than or equal to 1 kHz and less than or equal to 1 MHz.

14. A fluidic component, method, or reusable system as in any preceding claim, wherein the sensor comprises a plurality of sensors.

15. A fluidic component, method, or reusable system as in any preceding claim, wherein the sensor comprises two impendence detectors.

16. A fluidic component, method, or reusable system as in any preceding claim, wherein each impendence detector comprises seven or more electrodes.

17. A fluidic component, method, or reusable system as in any preceding claim, wherein the optical component comprises one or more paired light emitting diodes.

18. A fluidic component, method, or reusable system as in any preceding claim, wherein the optical component comprises two or more sources of electromagnetic radiation at different wavelengths.#14629375vl

Images