Evaluation of patients suffering from chronic pain

Abstract

A computer-implemented method for evaluation of a patient suffering from pain based on analysis of contextual data collected from the patient including conversational data taken from a patient interview. A machine learning model is trained to link the patient contextual data to one or more pain sentiment parameters to determine probability values for the pain sentiment parameters. The pain sentiment parameters relate to both current states of the patient, such as their current satisfaction with the pain treatment, and predicted future states of the patient, such as the likelihood of successful outcome for the patient with continuance of the current therapy unchanged or with continuance of the current therapy with specific modification. The pain sentiment parameters included in the machine learning model include patient states, patient therapy efficacy, patient therapy outcomes, patient therapy interventions. The patient contextual data includes conversational data collected from a patient interview.

Description

[0001] EVALUATION OF PATIENTS SUFFERING FROM CHRONIC PAIN

[0002] The invention relates to a computer-automated method for evaluation of patients suffering from chronic pain.

[0003] Chronic pain, such as back pain, is a common health condition. Back pain can be treated with a spinal cord stimulator (SCS) device or other implantable medical device (IMD). Treatment of back pain with SCS has varying degrees of success with different patients. While treatment with SCS produces a high degree of satisfaction in some patients, other patients find SCS unhelpful, while still further patients are unsure whether SCS is providing significant benefit. There has been extensive work on optimizing the control parameters used to define the electrical stimulation by SCS but this is not straightforward. The control parameter space is complex and feedback provided by patients is not straightforward to interpret and translate into changes of the control parameters.

[0004] The numerical rating scale for pain (NRS) is a subjective measure of pain provided by patient self-assessment. The NRS is widely used to evaluate patients suffering from chronic pain. NRS pain data is collected by asking a patient to state their pain intensity level on a scale of 0 and 10, where zero represents no pain and ten the worst pain imaginable. Another pain scale that is conceptually the same as the NRS is the visual analog scale (VAS). The success or failure of SCS therapy is often judged against NRS or VAS scores but this is a narrow measure of therapy efficacy, which does not capture a patient’s overall satisfaction with their chronic pain treatment. For example, some patients may be happier, i.e., more satisfied, when they continue an active lifestyle even if associated with higher pain levels than if they maintained a sedentary lifestyle, whereas other patients will have a straightforward correlation between happiness and pain intensity. Patient sentiment regarding their satisfaction with therapy is typically conveyed verbally in a conversation with a clinician and / or a caregiver. Presently there is no medical consensus on how to quantify patient sentiment. The relationship between patient sentiment on the one hand and pain state and the therapy efficacy on the other hand has not been extensively studied so remains largely unknown.

[0005] For example, a patient assessment is made after a trial use of SCS to decide whether the patient is a good candidate for an implanted SCS device. Similarly, patients with an implanted SCS device are assessed from time to time to evaluate the patient’s satisfaction with the therapy. In some cases, the device settings, i.e., the SCS control parameters, are recalibrated to attempt a re-optimization. In other cases, a decision to explant the device is made in case of patient dissatisfaction. It is possible to look at other metrics, not just NRS or VAS, when assessing patient satisfaction with their SCS therapy. These other metrics may include changes in disability, sleep and activity and the ability of the patient to set reasonable expectations and goals for managing their pain with SCS. However, there is no standard or consensus about what other metrics to consider or how they should be processed. Despite recalibrating an implanted SCS device from time to time by changing the SCS control parameters to re-optimize them, many patients and their clinicians conclude that the SCS device is not delivering effective pain relief and the SCS device is explanted.

[0006] One particular issue to take account of when treating patients with chronic pain therapy is to identify those patients who catastrophize their pain experience. Pain catastrophizing a maladaptive negative cognitive-affective response to anticipated or actual pain. Pain catastrophizing can be identified using the pain catastrophizing scale (PCS) or the coping strategies questionnaire (CSQ). Both these methods rely on a series of qualitative questions that can be time-consuming and imprecise to update frequently. Schutze et al., 2010 [1] developed a fear avoidance model to explain the observed association between pain catastrophizing and pain-related outcomes as disclosed by Quartana et al., 2009 [2], Pain catastrophizing inhibits a patient’s ability to sense an improvement in their pain experience, e.g., under SCS therapy, and therefore negatively impacts patient recovery and leads to deterioration of the patient’s pain experience. Currently, the adaptation of chronic pain treatment to avoid the onset of pain catastrophizing or to remove it is not well understood in general or in the specific context of SCS therapy.

[0007] Each patient’s pain therapy journey is a complex process influenced not only by pain intensity but more broadly by quality of life. Perceiving and recognizing improvements in the quality of life of a patient involves cognitive and affective processes, and currently there is no known solution to assess such processes in a quantitative or analytic manner. In summary, there is a need to provide quantitative metrics for pain treatment outcomes to improve upon known subjective measures of pain scoring using NRS or VAS, since this would improve long-term outcomes of patients, such as those utilizing SCS therapy.

[0008] According to first aspect of the disclosure there is provided a computer-implemented method for evaluation of a patient suffering from chronic pain, the method comprising: inputting patient contextual data collected from a patient suffering from (chronic) pain and / or other disease pattern or pathology that may require long-term treatment; processing the contextual data in a machine learning model trained to link patient contextual data of (chronic) pain patients and / or patients with any other disease pattern or pathology that may require long-term treatment to at least one pain sentiment parameter (or therapy sentiment parameter) so to determine a probability value for the at least one pain sentiment parameter; and outputting a probability value for the at least one pain sentiment parameter relating to one of a current state of the patient and a predicted future state of the patient, and / or outputting at least one of an inference and a prediction as a probability value of the at least one pain sentiment parameter, wherein the inference is a (probabilistic) characteristic of the current state of the patient and the prediction is a characteristic of a predicted future state of the patient.

[0009] The patient contextual data may include any combination of: conversational data collected from a patient interview.

[0010] In the following, the term "chronic pain" may include all other pains and / or disease pattern or pathology that may require long-term treatment. Patient contextual data can thus be used to support clinical decision-making and optimize therapy parameters while accounting for personal variations in the therapy response that is caused by factors other than pain experience itself, thereby increasing the likelihood of longterm therapy success. For example, recommendations may be made regarding pain management intervention for the patient, e.g., adjustment of SCS control parameters, suggest a clinical visit, and what external interventions to seek beyond the scope of the SCS or other specific physiological pain therapy, e.g., seek mindfulness, seek psychological therapy.

[0011] The pain sentiment parameters may include parameters selected from at least one of the group: patient states, patient therapy efficacy, patient therapy outcomes, and patient therapy interventions.

[0012] The at least one pain sentiment parameter output by the machine learning model may be selected from the group: probability of therapy success, probability of the patient feeling optimistic, probability of the patient suffering from pain catastrophizing; and probability of the patient needing an intervention.

[0013] The patient behavioral data collected from a device collecting data on patient behavior; and a patient self-assessed pain score according to a numerical rating scale for pain.

[0014] For processing the conversational data, a large language model and natural language processing may be used to extract a quantitative value of at least one conversation-derived variable selected from the group: sentiment score, stress level, cognitive load, cognitive bias, affective bias, and mood, said at least one conversation-derived variable being supplied to the machine learning model as input.

[0015] For processing the behavioral data, a dynamics model and kinematic processing may be used to extract a quantitative value of at least one behavioral variable selected from the group: posture, gait, sensory perception, and sensitivity, said at least one behavioral variable being supplied to the machine learning model as input. In one specific embodiment, a value of a pain sentiment parameter relating to a current state of the patient is combined with the patient’s self-assessed pain score to classify the patient into one of a plurality of patient outcome groups, said groups including at least a first group for patients who are dissatisfied with their pain therapy and a second group for patients who are satisfied with their pain therapy as well as optionally also at least a third group for patients who are neither satisfied nor dissatisfied with their pain therapy. The method may further comprise: providing a clinical recommendation for future treatment of the patient based on the probability value for the at least one pain sentiment parameter. For example, the patient being evaluated may be undergoing treatment with a spinal cord stimulation device (and / or another implantable medical device) and the clinical recommendation then comprises recommended values for a set of control parameters for the spinal cord stimulation device thereby allowing the spinal cord stimulation device to be adjusted for improved pain therapy.

[0016] In one specific embodiment, based on a value of a pain sentiment parameter relating to a future state of the patient a clinical recommendation is output regarding future treatment. For example, the patient being evaluated may be undergoing treatment with a spinal cord stimulation device (and / or another implantable medical device) and the clinical recommendation may be to re-calibrate the device or explant the device.

[0017] In certain embodiments, sentiment-pain inference is used to determine whether the problem is really with SCS or other external factors. Decision to explant the device may be one outcome.

[0018] In certain embodiments, the machine learning model incorporates a psychological model of pain that distinguishes between patients experiencing the sentiment of pain catastrophizing and those who are not.

[0019] A further aspect of the invention relates to a non-transitory computer readable medium storing instructions for performing a computer-implemented method as specified above.

[0020] A still further aspect of the invention relates to a computer apparatus having a memory and at least one processor, the memory having loaded therein instructions for performing a computer-implemented method as specified above, which, when executed by the at least one processor, cause the at least one processor to perform the computer-implemented method as specified above.

[0021] The computer apparatus may be remotely located from the SCS and / or the implantable medical device, for example in a distributed network and / or on / as a server (providing / serving the method as a cloud service).

[0022] This invention will now be further described, by way of example only, with reference to the accompanying drawings.

[0023] Figure 1 is a schematic diagram of an IMD with an SCS stimulator in a medical device communication system with standard architecture.

[0024] Figure 2 is a schematic diagram showing the SCS stimulator of Figure 1 attached to a lead arrangement of electrodes.

[0025] Figure 3 is a block diagram illustrating an example computing apparatus that may be used in the medical device communication system of Figure 1.

[0026] Figure 4 is a block diagram of a psychological model showing the effect of pain catastrophizing in patients with chronic pain.

[0027] Figure 5 is a block diagram of a data processing pipeline according to a first embodiment incorporating the chronic pain model of Figure 4.

[0028] Figure 6 is a block diagram of a data processing pipeline according to a second embodiment incorporating the chronic pain model of Figure 4.

[0029] Figure 7 is a block diagram of a data processing pipeline according to a third embodiment incorporating based on the chronic pain model of Figure 4. Figure 8 is a block diagram of a model for determining patient sentiment as a function of pain, affective state, and cognitive state.

[0030] Figure 9 is a model for determining patient pain state as a function of affective state, cognitive state and sentiment.

[0031] Figure 10 is a model for determining probabilities of different therapy outcomes as a function of pain, affective state, cognitive state and sentiment.

[0032] Figure 11 is a graph categorizing clinical outcomes by therapy outcome probabilities as a function of pain level.

[0033] Figure 12 is a graph of sentiment score against a numerical rating score of pain.

[0034] Figure 13 is a graph of the output of the model of Figure 10 plotting normalized likelihood ratio against numerical rating score to categorize patients between satisfied (S), dissatisfied (A) and borderline (T) .

[0035] In the following detailed description, for purposes of explanation and not limitation, specific details are set forth in order to provide a better understanding of the present disclosure. It will be apparent to one skilled in the art that the present disclosure may be practiced in other embodiments that depart from these specific details.

[0036] DEFINITIONS

[0037] Various psychological concepts as relevant for the present disclosure are defined as follows:

[0038] Experience a conscious event associated with an emotional and cognitive impact.

[0039] Affectivity / Affect the experience of feeling, emotion, or mood. Positive affectivity (PA) a measure of how much a person experiences positive affects (sensations, emotions, sentiments) leading to happiness, energy, enthusiasm etc.

[0040] Negative affectivity (NA) a measure of how much a person experiences negative affects (sensations, emotions, sentiments) leading to sadness, anxiety, depression etc.

[0041] Cognition the mental process of acquiring knowledge and understanding via thought, experience, and the senses.

[0042] Sentiment a persistent state of affective-conative attitude

[0043] Pain the experience of a physiological and a psychological response to a noxious stimulus.

[0044] Pain Catastrophizing a maladaptive negative cognitive-affective response to anticipated or actual pain, such as anxiety, fear, worry, magnification, rumination, and feelings of hopelessness.

[0045] In addition, we defined the following terms from language processing theory:

[0046] Contextual Data any information that can be used to characterize the situation of entities (i.e. whether a person, place or object) that are considered relevant to the interaction between a user and an application, including the user and the application themselves. Context is typically the location, identity and state of people, groups and computational and physical objects [3],

[0047] Decomposition Decomposition or decomposing behavioral data refers to segmenting data into unique packets, each carrying common information, not necessarily confined to languages. For example, video recording of a patient will have a segment where patient walks at specific time point, and at another time point, another segment will show the patient sitting. Each segment will contain different information.

[0048] Annotation for the purposes of the present document refers to the use of an algorithm to mark an audio file of a conversation between two or more persons to identify speaker boundaries (so-called speaker diarization) and thereby determine which segments of speech originate from which speaker.

[0049] Transcription refers to the conversion of speech, i.e., audio data, to text.

[0050] Dynamics Model For modeling the interactions among the variables of interest. Example includes gaiting model.

[0051] Kinematic Processing Process of extracting key features (numerical and categorical variables) derived from the dynamics model. Examples include walking characteristics such as speed and symmetry.

[0052] Large Language Model (LLM) a type of language model trained on massively large data sets based on artificial intelligence which is trained using self-supervised learning and semi-supervised learning.

[0053] Natural Language Processing (NLP) a type of processing of natural language using artificial intelligence to understand the meaning of speech including a speaker’s sentiment.

[0054] Machine Learning (ML) Model in the present document is used specifically to refer to ML models developed to connect conversational and behavioral data of a patient suffering from chronic pain with their pain therapy and therapy outcomes. The ML model may provide definitional, inferential or predictive links between contextual data of (chronic) pain patients (and / or patients with any other disease pattern or pathology that may require long-term treatment) and patient states, patient therapy efficacy, and patient therapy outcomes.

[0055] DESCRIPTION OF EMBODIMENTS

[0056] Those skilled in the art will further appreciate that the services, functions and steps explained herein may be implemented using software, i.e. a computer program, stored in memory and functioning in conjunction with a programmed microprocessor, or using an Application Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Programmable Logic Array (PLA), or a field programmable gate array (FPGA). As such references to a processor should include ASICs including artificial intelligence accelerator ASICs, DSPs, PLAs and FPGAs as well as central processor units (CPUs), graphics processor units (GPUs).

[0057] References to memory in the following may refer to any one or more of: a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), and a static random access memory (SRAM).

[0058] References to computer program in the following refer to machine readable program instructions for carrying out operations and may be assembler instructions, instruction-set- architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. It will be understood that features and elements of a standard architecture described above with reference to Figure 1 will be incorporated in embodiments of the invention as appropriate and specific detail already described above may not be repeated in the following detailed description for corresponding features.

[0059] An SCS device is a type of IMD that is used for chronic pain therapy of a patient. An SCS device uses electrical current sent through electrodes on one or more leads implanted in the epidural space dorsal to the spinal cord in an SCS device. Control parameters include at least: pulse width, frequency, amplitude and electrode selection. Additionally, duty cycling and burst stimulation can be introduced as further control parameters to modulate therapy and better manage battery consumption through the introduction of alternating periods of carrier and envelope stimulation.

[0060] An IMD typically forms part of a medical device communication system (MDCS) that is configured to upload clinical data from the IMD on a continual basis via a MDCS to a remotely located data repository, which may be a data center. In an SCS device, clinical data includes both patient data relating to physiological monitoring of a patient's health, for example as collected by sensors of the IMD, and device data related to status and operation of the IMD, for example data logging each time an SCS device is used and at what stimulation level in terms of electrical pulse width and pulse frequency as well as control signals issued to the SCS device to optimize delivery of pain relief. These clinical data can then be accessed by health care staff accessing the data center using suitably designed software applications. Treatment plans and, if necessary, interventions can then be decided upon based on analysis of these clinical data. The analysis may be expert analysis by a health care professional or computer-automated analysis with a software program, e.g., running on computing resource at a neuro-service data center, or a combination of both.

[0061] Figure 1 shows a standard architecture of a MDCS 1 for communication between an IMD 10 implanted in a patient and a remotely located data center 60, which is accessible to health care staff using suitable application software. The IMD 10 includes therapeutic components for patient treatment, such as an SCS device 11 (or another IMD stimulator) incorporating an implantable pulse generator (IPG), and therapeutic components for patient monitoring, such as an IMD sensor 13. The IMD 10 further comprises chipsets providing computing and telecommunications resource in the form of memory 17, a processor 18 and a wireless personal area network (WPAN) transceiver 12. The IMD 10 is powered by an electrical energy source 14, in this example by a rechargeable battery 14. The IMD's rechargeable battery 14 of an implanted IMD can be charged in a contactless manner by a charger 20, for example a resonant inductive charger, which is placed on the patient's skin adjacent the implanted IMD 10 to charge its rechargeable battery 14. The charger 20 is itself provided with an energy source 24, here a rechargeable battery 24, as well as an external mains power connection 29, for example an external power jack, to power the charger 20 so that its rechargeable battery 24 can be recharged. Alternatively, the charger's battery 24 can be recharged wirelessly by placing the charger 20 on a mains-powered charging pad. The IMD WPAN transceiver 12 uses a suitable WPAN protocol such as Bluetooth Low Energy (BLE), Medical Implant Communication System (MICS) or Medical Device Radiocommunications Service (MedRadio), the latter two being almost identical protocols.

[0062] The patient is provided with a patient remote controller 30 (hereinafter also called “patient remote 30”), for example a smartphone with wireless transceivers 32, 35, 36 respectively for WPAN (e.g. BLE), LPWAN, cellular (e.g., 4G / LTE / 5G) and WLAN communication. If a smartphone is used, this is a smartphone that is possessed, e.g., owned, by the patient in which the IMD 10 is implanted and on which a software application ('app') is installed. The app is then a so-called Software as a Medical Device (SaMD) which is defined by the United States Food and Drug Administration (FDA) as software intended to be used for one or more medical purposes that perform these purposes without being part of a hardware medical device.

[0063] The cellular transceiver 35 and WLAN transceiver 36 provide two different data communication paths for the patient remote 30 to upload clinical data from the IMD 10 via an internet connection 50 to a remotely located data center 60 (hereinafter also called “backend 60”) acting as repository for storage of clinical data and / or as a host for the services, for example a neuro-service data center. The remote's WLAN transceiver 36 can upload data to the backend 60 via a router 38 and a telephone line 40 (or telephone network 40) using a wired internet connection 50. The internet connection 50 may provide access to one or more distributed networks and / or cloud services. The remote's cellular transceiver 35 can upload data to the backend 60 via one or more cellular network base stations 42 (cellular towers), for example LPWAN-capable cellular network base station. In some cases, instead of a public communication network, a dedicated point-to-point transmission, e.g., via a dedicated telephone line, may be provided for uploading clinical data. In all these scenarios, uploading of clinical data from the IMD 10 to the backend 60 takes place via the intermediary of the patient remote 30, the latter thereby acting as a relay device.

[0064] Health care staff, such as health care professionals (HCPs), clinical specialists and representatives and remote care team members have access to the backend 60 via suitable portals 70 with the aid of a software application running on the backend 60 and / or the portal 70 to provide the necessary user interfacing, diagnostics and so forth. At least one portal 70 is provided by at least one workstation for health care staff to access a data center, e.g., the backend 60. Analysis and diagnostic software may also be run at the backend 60 to analyze clinical data from individual patients or groups of patients.

[0065] Figure 2 is a schematic diagram showing the SCS device 11 of the IMD 10 attached to a lead arrangement 80 comprising one or more leads 82 with each lead incorporating one or more electrodes 84. In the illustrated example, there is one electrode per lead and the electrodes 84 are N in number and labeled, El, E2, E3 .... EN. The SCS device 11 comprises an electrode drive circuit 19 for providing a suitable drive current to each of the electrodes 84 according to a set of control parameters that follow a treatment plan as controlled by a control circuit 16. The control circuit 16 delivers control signals to the electrode drive circuit 19 according to a treatment plan devised by a computer program running on the processor 18 (e.g., a microprocessor) of the IMD 10. The computer program is stored in the IMD memory 17. These hardware and software components operate collectively to provide an intelligent pulse generator to deliver electrical pulses to the electrodes conforming to a particular set of parameters, including amplitude, pulse width, frequency and / or duty cycle to provide stimulation therapy. For SCS, the electrodes are implanted at or near a patient’s spinal cord to direct electrical signals into the patient’s tissue for spinal cord stimulation. The SCS device is implanted subcutaneously. Figure 3 is a block diagram illustrating an example computing apparatus 100 that may be used in connection with various embodiments described herein. For example, computing apparatus 100 may be used as the remote computing resource at the neuro-service data center 60 of Figure 1.

[0066] Computing apparatus 100 can be a server or any conventional personal computer, or any other processor-enabled device that is capable of wired or wireless data communication. Other computing apparatus, systems and / or architectures may be also used, including devices that are not capable of wired or wireless data communication, as will be clear to those skilled in the art.

[0067] Computing apparatus 100 preferably includes one or more processors, such as processor 110. The processor 110 may be for example a CPU, GPU, TPU or arrays or combinations thereof such as CPU and TPU combinations or CPU and GPU combinations. Additional processors may be provided, such as an auxiliary processor to manage input / output, an auxiliary processor to perform floating point mathematical operations (e.g. a TPU), a specialpurpose microprocessor having an architecture suitable for fast execution of signal processing algorithms (e.g., digital signal processor, image processor), a slave processor subordinate to the main processing system (e.g., back-end processor), an additional microprocessor or controller for dual or multiple processor systems, or a coprocessor. Such auxiliary processors may be discrete processors or may be integrated with the processor 110. Examples of CPUs which may be used with computing apparatus 100 are those available from Intel Corporation of Santa Clara, California or Nvidia Corporation, Santa Clara, California.

[0068] Processor 110 is connected to a communication bus 105. Communication bus 105 may include a data channel for facilitating information transfer between storage and other peripheral components of computing apparatus 100. Communication bus 105 further may provide a set of signals used for communication with processor 110, including a data bus, address bus, and control bus (not shown). Communication bus 105 may comprise any standard or non-standard bus architecture such as, for example, bus architectures compliant with industry standard architecture (ISA), extended industry standard architecture (EISA), Micro Channel Architecture (MCA), peripheral component interconnect (PCI) local bus, or standards promulgated by the Institute of Electrical and Electronics Engineers (IEEE) including IEEE 488 general-purpose interface bus (GPIB), IEEE 696 / S-100, and the like.

[0069] Computing apparatus 100 preferably includes a main memory 115 and may also include a secondary memory 120. Main memory 115 provides storage of instructions and data for programs executing on processor 110, such as one or more of the functions and / or modules discussed above. It should be understood that computer readable program instructions stored in the memory and executed by processor 110 may be assembler instructions, instructionset-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, configuration data for integrated circuitry, or either source code or object code written in and / or compiled from any combination of one or more programming languages, including without limitation Smalltalk, C / C++, Java, JavaScript, Perl, Visual Basic, .NET, and the like. Main memory 115 is typically semiconductor-based memory such as dynamic random access memory (DRAM) and / or static random access memory (SRAM). Other semiconductor-based memory types include, for example, synchronous dynamic random access memory (SDRAM), Rambus dynamic random access memory (RDRAM), ferroelectric random access memory (FRAM), and the like, including read only memory (ROM).

[0070] The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0071] Secondary memory 120 may optionally include an internal memory 125 and / or a removable medium 130. Removable medium 130 is read from and / or written to in any well-known manner. Removable storage medium 130 may be, for example, a magnetic tape drive, a compact disc (CD) drive, a digital versatile disc (DVD) drive, other optical drive, a flash memory drive, etc.

[0072] Removable storage medium 130 is a non-transitory computer-readable medium having stored thereon computer-executable code (i.e., software) and / or data. The computer software or data stored on removable storage medium 130 is read into computing apparatus 100 for execution by processor 110.

[0073] The secondary memory 120 may include other similar elements for allowing computer programs or other data or instructions to be loaded into computing apparatus 100. Such means may include, for example, an external storage medium 145 and a communication interface 140, which allows software and data to be transferred from external storage medium 145 to computing apparatus 100. Examples of external storage medium 145 may include an external hard disk drive, an external optical drive, an external magneto- optical drive, etc. Other examples of secondary memory 120 may include semiconductor- based memory such as programmable read-only memory (PROM), erasable programmable readonly memory (EPROM), electrically erasable read-only memory (EEPROM), or flash memory (block-oriented memory similar to EEPROM).

[0074] As mentioned above, computing apparatus 100 may include a communication interface 140. Communication interface 140 allows software and data to be transferred between computing apparatus 100 and external devices (e.g., printers), networks, or other information sources. For example, computer software or executable code may be transferred to computing apparatus 100 from a network server via communication interface 140. Examples of communication interface 140 include a built-in network adapter, network interface card (NIC), Personal Computer Memory Card International Association (PCMCIA) network card, card bus network adapter, wireless network adapter, Universal Serial Bus (USB) network adapter, modem, a network interface card (NIC), a wireless data card, a communications port, an infrared interface, an IEEE 1394 fire-wire, or any other device capable of interfacing system with a network or another computing device. Communication interface 140 preferably implements industry-promulgated protocol standards, such as Ethernet IEEE 802 standards, Fiber Channel, digital subscriber line (DSL), asynchronous digital subscriber line (ADSL), frame relay, asynchronous transfer mode (ATM), integrated digital services network (ISDN), personal communications services (PCS), transmission control protocol / Intemet protocol (TCP / IP), serial line Internet protocol / point to point protocol (SLIP / PPP), and so on, but may also implement customized or non-standard interface protocols as well.

[0075] Software and data transferred via communication interface 140 are generally in the form of electrical communication signals 155. These signals 155 may be provided to communication interface 140 via a communication channel 150. In an embodiment, communication channel 150 may be a wired or wireless network, or any variety of other communication links. Communication channel 150 carries signals 155 and can be implemented using a variety of wired or wireless communication means including wire or cable, fiber optics, conventional phone line, cellular phone link, wireless data communication link, radio frequency ("RF") link, or infrared link, just to name a few.

[0076] Computer-executable code (i.e., computer programs or software) is stored in main memory 115 and / or the secondary memory 120. Computer programs can also be received via communication interface 140 and stored in main memory 115 and / or secondary memory 120. Such computer programs, when executed, enable computing apparatus 100 to perform the various functions of the disclosed embodiments as described elsewhere herein.

[0077] In this disclosure, the term "computer-readable medium" is used to refer to any non- transitory computer-readable storage media used to provide computer-executable code (e.g., software and computer programs) to computing apparatus 100. Examples of such media include main memory 115, secondary memory 120 (including internal memory 125, removable medium 130, and external storage medium 145), and any peripheral device communicatively coupled with communication interface 140 (including a network information server or other network device). These non-transitory computer-readable media are means for providing executable code, programming instructions, and software to computing apparatus 100. In an embodiment that is implemented using software, the software may be stored on a computer-readable medium and loaded into computing apparatus 100 by way of removable medium 130, I / O interface 135, or communication interface 140. In such an embodiment, the software is loaded into computing apparatus 100 in the form of electrical communication signals 155. The software, when executed by processor 110, preferably causes processor 110 to perform the features and functions described elsewhere herein.

[0078] I / O interface 135 provides an interface between one or more components of computing apparatus 100 and one or more input and / or output devices. Example input devices include, without limitation, keyboards, touch screens or other touch-sensitive devices, biometric sensing devices, computer mice, trackballs, pen-based pointing devices, and the like. Examples of output devices include, without limitation, cathode ray tubes (CRTs), plasma displays, light-emitting diode (LED) displays, liquid crystal displays (LCDs), printers, vacuum florescent displays (VFDs), surface-conduction electron-emitter displays (SEDs), field emission displays (FEDs), and the like.

[0079] Computing apparatus 100 also includes optional wireless communication components that facilitate wireless communication over a voice network and / or a data network. The wireless communication components comprise an antenna system 170, a radio system 165, and a baseband system 160. In computing apparatus 100, radio frequency (RF) signals are transmitted and received over the air by antenna system 170 under the management of radio system 165.

[0080] Antenna system 170 may comprise one or more antennae and one or more multiplexors (not shown) that perform a switching function to provide antenna system 170 with transmit and receive signal paths. In the receive path, received RF signals can be coupled from a multiplexor to a low noise amplifier (not shown) that amplifies the received RF signal and sends the amplified signal to radio system 165.

[0081] Radio system 165 may comprise one or more radios that are configured to communicate over various frequencies. In an embodiment, radio system 165 may combine a demodulator (not shown) and modulator (not shown) in one integrated circuit (IC). The demodulator and modulator can also be separate components. In the incoming path, the demodulator strips away the RF carrier signal leaving a baseband receive audio signal, which is sent from radio system 165 to baseband system 160.

[0082] If the received signal contains audio information, then baseband system 160 decodes the signal and converts it to an analogue signal. Then the signal is amplified and sent to a speaker. Baseband system 160 also receives analogue audio signals from a microphone. These analogue audio signals are converted to digital signals and encoded by baseband system 160. Baseband system 160 also codes the digital signals for transmission and generates a baseband transmit audio signal that is routed to the modulator portion of radio system 165. The modulator mixes the baseband transmit audio signal with an RF carrier signal generating an RF transmit signal that is routed to antenna system 170 and may pass through a power amplifier (not shown). The power amplifier amplifies the RF transmit signal and routes it to antenna system 170 where the signal is switched to the antenna port for transmission.

[0083] Baseband system 160 is also communicatively coupled with processor 110, which may be a central processing unit (CPU). Processor 110 has access to data storage areas 115 and 120. Processor 110 is preferably configured to execute instructions (i.e., computer programs or software) that can be stored in main memory 115 or secondary memory 120. Computer programs can also be received from baseband system 160 and stored in main memory 110 or in secondary memory 120 or executed upon receipt. Such computer programs, when executed, enable computing apparatus 100 to perform the various functions of the disclosed embodiments. For example, data storage areas 115 or 120 may include various software modules.

[0084] The computing apparatus further comprises a display 175 directly attached to the communication bus 105 which may be provided instead of or addition to any display connected to the VO interface 135 referred to above.

[0085] Various embodiments may also be implemented primarily in hardware using, for example, components such as application specific integrated circuits (ASICs), programmable logic arrays (PLA), or field programmable gate arrays (FPGAs). Implementation of a hardware state machine capable of performing the functions described herein will also be apparent to those skilled in the relevant art. Various embodiments may also be implemented using a combination of both hardware and software.

[0086] Furthermore, those of skill in the art will appreciate that the various illustrative logical blocks, modules, circuits, and method steps described in connection with the abovedescribed figures and the embodiments disclosed herein can often be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled persons can implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the invention. In addition, the grouping of functions within a module, block, circuit, or step is for ease of description. Specific functions or steps can be moved from one module, block, or circuit to another without departing from the invention.

[0087] Moreover, the various illustrative logical blocks, modules, functions, and methods described in connection with the embodiments disclosed herein can be implemented or performed with a general-purpose processor, a digital signal processor (DSP), an ASIC, FPGA, or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general- purpose processor can be a microprocessor, but in the alternative, the processor can be any processor, controller, microcontroller, or state machine. A processor can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.

[0088] Additionally, the steps of a method or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium including a network storage medium. An exemplary storage medium can be coupled to the processor such that the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium can be integral to the processor. The processor and the storage medium can also reside in an ASIC.

[0089] A computer readable storage medium, as referred to herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.

[0090] Any of the software components described herein may take a variety of forms. For example, a component may be a stand-alone software package, or it may be a software package incorporated as a "tool" in a larger software product. It may be downloadable from a network, for example, a website, as a stand-alone product or as an add-in package for installation in an existing software application. It may also be available as a client- server software application, as a web-enabled software application, and / or as a mobile application.

[0091] Embodiments of the present invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer readable program instructions.

[0092] The computer readable program instructions may be provided to a processor of a general- purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and / or other devices to function in a particular manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function / act specified in the flowchart and / or block diagram block or blocks.

[0093] The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0094] The illustrated flowcharts and block diagrams illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the blocks may occur out of the order noted in the Figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustration, and combinations of blocks in the block diagrams and / or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.

[0095] Figure 4 is block diagram of a psychological model showing the effect of pain catastrophizing in patients with chronic pain showing the relationship between pain experience and its response. Pain score and sentiment are example means of measurement for the pain experience and affective responses, respectively. This fear avoidance model is a revised version of that disclosed by Schiitze et al., 2010 [1], As illustrated in Figure 4, according to this model of pain, sentiment 201 includes negative affectivity, e.g. pain catastrophizing 204 and positive affectivity, e.g. not pain catastrophizing or no fear of pain 205, which are negative and positive affectivities respectively. Pain 201 includes an experience, e.g. pain experience 207 and may be characterized by a pain score 208, e.g., an NRS score. The sentiment of pain catastrophizing 204 causes deterioration 212 of sentiment feeding back negatively to pain 200, as may be measured by an NRS score, whereas a patient not suffering from fear of pain 205 has positive feedback of improvement 210 to pain 200. The patient’s pain-related outcome diverges to either deterioration or improvement depending on the patient’s attitude towards pain. If a patient has no fear towards pain, the patient can confront their pain and work towards achieving successful pain therapy. On the other hand, if a patient has or develops a fear of pain, the patient may withdraw from engaging in pain therapies, which exacerbates the pain especially when reinforced with negative emotions.

[0096] In the following, we describe first, second and third example computer-automated processes based on the chronic pain model of Figure 4. These embodiments construct a mathematical relationship between a patient’s sentiment towards their pain therapy (e.g. SCS), where the sentiment is extracted from contextual data, which may include conversational data (e.g. audio, video, text) and / or behavioral data (e.g. IMD sensor data), and optionally also selfreported pain scores (e.g., NRS). Each embodiment processes raw conversational data and / or raw behavioral data captured from a patient to determine patient parameters relevant for diagnosis and treatment, such as the patient’s pain state, probabilistic predictions of treatment success with different treatment types, e.g., through physiological therapy with a medical device or chirurgical intervention, or through psychological therapies such as counseling or well-being activities. In this way, interventions can be selected that are personalized to the patient.

[0097] Patient conversational data is analyzed using computer-automated processes which make combined use of a large language model (LLM), natural language processing (NLP) and a machine learning (ML) model. In this way, sentiments are extracted from a patient interview. The patient interview may be, for example, a conversation between a patient and a health care professional or a robot acting in the role of a health care professional. Conversational data will include audio but may optionally also include video, e.g., to correlate visual cues of the patient, such as facial expression or posture, with speech.

[0098] Patient behavioral data is analyzed using computer-automated processes which make combined use of a dynamics model and kinetic processing. Patient behavioral data may be data collected from a device tracking patient behavior, such as video data from a camera observing the patient, or data collected from a device carried by or implanted in a patient, such as an IMD, a step counter, an electrocardiogram (ECG), a smart watch etc. For example, behavioral data may be indicative of patient activity and exercising as well as sleep. (Behavioral data is defined in this disclosure as not including conversational data.)

[0099] Figure 5 is a block diagram of a data processing pipeline and associated method according to a first embodiment incorporating the chronic pain model of Figure 4 which processes a patient’s raw conversational data 214 to infer the patient’s pain states and predict therapy outcomes of different possible therapies, e.g. inference and prediction 228. Conversational data 214 may be in the form of video, audio, and / or text recordings. Raw conversational data 214 is collected by patient interview, e.g., during a meeting between a patient and a health care professional. Multiple sources of the recordings may be combined. To pre-process the raw conversational data 214, transcription 216 is performed, e.g., to convert video and / or audio data to text. In case of conversations involving multiple speakers, annotation 218 is performed to tag speaker identity (speaker diarization).

[0100] The transcribed and annotated conversational data is then input into a large language model (LLM) 222, which is used to summarize and extract information from the pre-processed conversational data. The LLM 222 may be configured to extract features from the conversational data including quantitative or categorical variables such as:

[0101] • Sentiment score

[0102] • Stress level

[0103] • Cognitive load

[0104] • Cognitive bias Affective bias

[0105] Mood (depression / anxiety level)

[0106] The input prompt to the LLM 222 is engineered to obtain either or both information extraction or / and summaries of the conversational data. Zero-shot, one-shot, and / or few-shot prompting may be utilized depending on the complexity of the inputs and outputs. The LLM may be fine-tuned to interpret chronic-pain-specific contexts more accurately and precisely.

[0107] The output of the LLM 222 is then input into a natural language processing (NLP) 224 (hereinafter also called “NLP algorithm 224"), which quantifies and extracts features. The output of the NLP algorithm 224 is input into a bespoke machine learning (ML) model 226 (of pain state from patient data) that has been trained to link the patient contextual data to one or more pain sentiment parameters to determine probability values for the pain sentiment parameters. The pain sentiment parameters relate to both current states of the patient, such as their current satisfaction with the pain treatment, and predicted future states of the patient, such as the likelihood of successful outcome for the patient with continuance of the current therapy unchanged or with continuance of the current therapy with specific modification. The pain sentiment parameters included in the machine learning model include patient states, patient therapy efficacy, patient therapy outcomes, patient therapy interventions. The ML model 226 maps the relationships between the patients’ pain, affective, and cognitive states, and sentiment. The ML model 226 then outputs inferences for the pain, psychological, and physiological health states of the patient and the probability of success of different interventions, for example one or more physiological therapies, psychological therapies and surgeries. The likely success for different interventions is thus predicted.

[0108] Example pain sentiment parameters that may be output from the ML model 226 as predictive indicators (inference and prediction) 228 include:

[0109] 1. Probability of therapy success

[0110] 2. Probability of feeling optimistic

[0111] 3. Probability of patient suffering from pain catastrophizing

[0112] 4. Probability of needing an intervention By way of example, these outputs from the ML model 226 may be used for diagnosis and treatment as follows.

[0113] 1. Therapy success: If probability of successful therapy is below a threshold (or probability of unsuccessful therapy is above a threshold), this may trigger a clinician intervention and also trigger the ML model to recommend other therapies with a higher probability of success. The ML model may, for example, generate an output that ranks different therapies by probability of success, these outputs being individual to the patient whose conversational and / or behavioral data is being processed.

[0114] 2. Feeling Optimistic: The probability of feeling optimistic may serve as an indicator to assess the patients’ capability to confront pain and work towards therapy success.

[0115] 3. Pain Catastrophizing: The probability of a patient suffering from pain catastrophizing may trigger a psychological therapy.

[0116] 4. Needing Intervention: The probability of needing an intervention may, for example, calculate any of the following probabilities:

[0117] • checking on the patient by a caregiver,

[0118] • making an appointment with the treating clinician,

[0119] • recommending medications,

[0120] • recommending activity increase / decrease,

[0121] • recommending social interaction.

[0122] The ML model parameters may be obtained for the general population, for subpopulations and for individual patients. The ML model parameters may be used to cluster patient to different patient phenotypes.

[0123] Training data materials for the ML model may include any combination of the following: • Conversational data and / or behavioral data between the patients and their clinicians, caregivers, and others who are involved in the pain experience

[0124] • Research literature documents

[0125] • Device / operational manuals

[0126] • Treatment usage / progress

[0127] • Expert reports

[0128] • Patient demographics data

[0129] Summaries for both a single patient interaction or multiple interactions with the same patient may be used. A single interaction summary has more temporal precision but is prone to be noisier in that it may more likely be influenced by factors other than pain-related ones. Multiple interactions with the patient (e.g. first meeting and follow-ups) may be summarized together to obtain event-triggered therapy efficacy. Patient interactions over a short time period (e.g. within a three-month period) and patient interactions over a long time period (e.g. over a greater time than three months) may be summarized together to obtain general therapy efficacy.

[0130] In summary, qualitative contextual information from a conversation between the patient and a medical professional (e.g., clinician, caregiver) is extracted from the raw conversational data and quantified using a processing pipeline of transcription, annotation, language interpretation with LLM and NLP followed by assessment with a ML model to determine probabilities of different pain therapy outcomes.

[0131] Figure 6 is a block diagram of a data processing pipeline and associated method according to a second embodiment incorporating the chronic pain model of Figure 4. The second embodiment is the same as the first embodiment apart from the addition of diction assessment 230. Namely, the second embodiment, like the first embodiment, processes a patient’s raw conversational data 214 to infer the patient’s pain states and predict therapy outcomes of different possible therapies 228 using a processing pipeline of transcription 216 and annotation 218 followed by LLM processing 222, NLP 224 and passing through a ML model 226 trained to convert the patient conversational data as processed by the LLM and NLP into probabilistic values of one or more pain sentiment parameters relating to current or future patient states. In addition to the linguistic components in the data being analyzed by the LLM and NLP as in the first embodiment, there is analysis of metadata such as articulation to make a diction assessment 230. Diction assessment is used to capture neurological and / or psychological characteristics of the patient such as vocal stress, cognitive load, cognitive bias, and affective bias. In a variant, the stress may be deduced from video data from the facial expression of the patient, either with or without the audio data.

[0132] Figure 7 is a block diagram of a data processing pipeline and associated method according to a third embodiment incorporating the chronic pain model of Figure 4. The patient data input for the third embodiment is behavioral data 215 instead of conversational data. Behavioral data may be in the form of video data or sensor data, e.g., collected from sensors carried by or implanted in the patient in an IMD. Behavioral data may include body gesture data, patient movement data, patient posture data, patient gait data, patient sensory perception data and patient sensitivity data. The behavioral data 215 is pre-processed initially by decomposition 220 and annotation 218 followed by a dynamics model 232 for information extraction and summary, and then kinematic processing 234 for quantitative analysis and feature engineering. The output of the kinematic processing 234 is input to a ML model 226 trained to convert the patient behavioral data 215 as processed by the dynamics model and kinematic processing into probabilistic values of one or more pain sentiment parameters relating to current or future patient states.

[0133] Decomposition here may encompass greater use than just languages. Behavioral data may be broken down to several unique segments containing different information. For example, walking video contains visual (who and where) and auditory information as well as posture and movement data. These data need to be decomposed. On the other hand, conversational data is mostly auditory (time-series data). Here once we annotate which portions are relevant, we may rely on traditional linguistic decomposition (LLM).

[0134] It will be understood that the third embodiment may be combined with the first or second embodiments so that both conversational data and behavioral data are jointly analyzed. The ML models developed for embodiments of the present invention exist on three levels: (1) definitional, (2) inferential, and (3) predictive.

[0135] 1 Definitional to model the underlying mechanisms why different patients respond to pain experience differently as defined by the relationship between a patient’s pain, affective states, cognitive states and contextual sentiment.

[0136] 2. Inferential to estimate the patients’ pain state as a function of patients’ sentiment towards the pain and therapy experience.

[0137] 3 Predictive to estimate the probability of therapy success and indications related to the therapy outcomes.

[0138] These ML models can be used to guide optimization of therapy settings and to inform intervention-related decision-making. Patient conversational and behavioral data include speech data and metadata such as stress assessment, speech cadence, cognitive load.

[0139] These ML models thus evaluate therapy satisfaction and can guide therapy and provide insights into treatment efficacy.

[0140] In the above embodiments, the probabilistic predictions of treatment success may be limited to the treatment which the patient is currently being given, e.g. SCS with an IMD, and then may be used to adjust the treatment by adjusting the control parameters used by the IMD to improve predicted therapy efficacy and hence increase the likelihood of therapy success. The probabilistic predictions of treatment success may also be used to assist a clinician in making decisions on whether to continue or abandon the current therapy and which alternate therapy would be the most likely to be the most successful. For example, the method provides a way to quantitatively assess patient pain state and SCS satisfaction to evaluate a trial outcome and to optimize care post implant, thereby providing stable therapeutic efficacy over the long term by SCS. Figure 8 is a block diagram of a model (Model 1) for determining patient sentiment 201 towards their pain experience as a function of pain 200, cognitive state 202 and affective state 203. Pain states are obtained from conversational data, for example by patients reporting on subjective pain levels on the NRS or VAS scale, or extracted from contextual data. Sentiment is obtained from conversational data directly by patients or is extracted from contextual data. Cognitive states (mindfulness, goal-setting, etc.) are obtained from conversational data, e.g., as directly reported by patients, or extracted from contextual data. Affective states (fear, optimism, etc.) are obtained from conversational data, e.g., as directly reported by patients, or extracted from contextual data.

[0141] One example ML model implementation uses logistic regression to determine sentiment according to the formula: where

[0142] ( / ) is a cumulative Gaussian function

[0143] Aindicates estimated parameters i indicates ithvariable (i.e. pain, cognition, affectivity)

[0144] A is the height of the sigmoidal function

[0145] So is the lower plateau

[0146] E is an error term.

[0147] Instead of a Gaussian function, a Weibull function may be used to account for asymmetry.

[0148] The function may be estimated for singular input parameter, e.g., or for multiple parameters, either independently or in composite variables where other variables may exist in non-sigmoidal function such as nonlinear, discontinuous gating function. The sentiment function can be obtained on a whole population level, sub- population level or individual level as desired. A whole population model provides the criteria guidelines for defining successful therapy in general, whereas a personalized individual model can account for individual differences in utilizing cognitive and affective functions to achieve a target sentiment that subsequently impacts therapy outcomes.

[0149] Figure 9 is a model (Model 2) for determining patient pain state 209 as a function of cognitive state 202, affective state 203 and sentiment 201. Pain state is a function of sentiment given cognitive and affective states (Bayesian estimate).

[0150] In the top model (pointing to the sentiment), sentiment is mediating the effect of cognition / affectivity on the pain state. In the bottom model (pointing to pain state), the effect of cognition / affectivity is approximated to have direct effect on the pain state. In short, h' is an approximation, potentially ignoring the mediating effect of the sentiment.

[0151] The pain state function may be approximated as a function of sentiment, cognitive, and affective states (maximum-likelihood estimate). The ML model may be implemented with nonlinear regression using an inverse cumulative Gaussian function as the target function, for example as follows:

[0152] Pain may be approximated by singular or multiple-variable function, with estimated variables forming multi-dimensional vectors. Another option for implementing the ML model is with linear regression, for example as follows: pain = bs■ sentiment + bc■ cog + ba■ aff + b0+ epain

[0153] A still further option for implementing the ML model is to combine multiple functions, for example a nonlinear regression function (e.g. inverse cumulative Gaussian), a discontinuous gating function, and / or a linear regression function. Figure 10 is a model (Model 3) for determining probabilities, Pr, of different therapy outcomes as a function of pain 200, cognitive state 202, affective state 203, and sentiment 201. The outcome indication provided by the probability may be for example: success, optimism, pain catastrophizing, or intervention as described further above. The probability of therapy success, Pr success, is expressed as a function of sentiment given pain, affective, and cognitive states and may be approximated as a function of optimism probability. The probability of optimism, Pr optimism, towards pain and therapy experience is expressed as a function of sentiment given pain, affective, and cognitive states. The optimism function may provide indication for the patient’ s willingness to confront the pain experience and work towards achieving their goals of optimal therapy outcomes. The probability of pain catastrophizing, Pr catastrophizing, is expressed as a function of sentiment given pain, affective, and cognitive states. This function is inversely related to the therapy success probability, with other factors accounted for in the equation by the factor, a. When the probability, Pr catastrophizing, exceeds a certain threshold, an alert may be generated and sent to the relevant care providers to prompt an intervention with the aim of avoiding further development towards a pain catastrophizing patient state.

[0154] One example ML model implementation is logistic regression to estimate the probability: where y is the outcome indication variable

[0155] Another example ML model implementation is a linear regression of the outcome indication variable as an approximation:

[0156] A still further example ML model implementation is a combination of logistic, discontinuous gating function, and / or linear regressions. Yet another ML model implementation is a form of likelihood ratio function such as log-likelihood ratio and normalized likelihood ratio between different probability functions defined above.

[0157] The outcome indication variable in the ML model may be for example: success, optimism, pain catastrophizing, or intervention as described further above.

[0158] Figure 11 is a graph categorizing clinical outcomes by therapy outcome probabilities as a function of pain level. Here the pain-sentiment space output by the ML model is used to classify patient outcomes into one of a plurality of clinical outcome zones for pain therapy which are divided from each other by thresholds that may be set on an individual patient level. Patients are thus grouped by these outcome zones with each individual patient being classified into one of these outcome zones. By way of example, Figure 11 illustrates a classification into three outcome zones. In general, the number of outcome zones may be 2 or more. The three outcome zones in this example are:

[0159] Satisfaction zone', patient achieves both low pain and positive sentiment (satisfied patient).

[0160] Adverse zone. patient has high pain and negative sentiment (dissatisfied patient). Transition zone : patient achieves moderate pain level and neutral sentiment in between the satisfaction and adverse zones (borderline patient).

[0161] The pain and sentiment values may be direct measures, i.e., these parameters themselves, or derived measures from the model described with reference to Figure 10. One example ML model implementation provides a probability associated with each classes (outcome zones). Another example ML model implementation is binary classification model with 2 outcome zones, satisfaction zone and non-satisfaction zone (neutral & adverse zones). A still further example ML model implementation divides zones with Tp,s (satisfactory pain threshold), Tp,a (adverse pain threshold), Ts,s (satisfactory sentiment threshold), and Ts,a (adverse sentiment threshold). These classifications may be made on either a population level or an individual patient level.

[0162] The ML model may use multiple-logistic regression with probability and class outputs. An alternative is for the ML model to use a tree-based classification model with probability and class outputs.

[0163] Figure 12 is a graph of sentiment score against numerical rating score, NRS, which is used to define three example satisfaction zones based on the model of Figure 8. Sentiment as a function of pain score was obtained using data from a clinical study. Filled circles are patient data points, and the solid curve is the best logistic function fit. The curve provides a quantified definition of the relationship between pain-related outcomes and sentiment, which can be used as the basis for inference and predictive modeling.

[0164] Figure 13 is a graph of the output of the model of Figure 10 plotting normalized likelihood ratio, NLR, against numerical rating score, NRS, to categorize patients between satisfied (S), dissatisfied (A) and borderline (T). Filled circles are data from the same clinical study as used for Figure 12. The solid curve is the best fit to the normalized likelihood ratio between Pr optimism and Pr catastrophizing. In this example, the satisfaction zone (S), transition zone (T), and adverse zone (A) were determined based on the population Pr optimism prediction. In other implementations, zone boundaries can be found on an individual level after collecting a patient’s feedback. REFERENCES

[0165] 1. Schiitze R, Rees C, Preece M, Schiitze M. Low mindfulness predicts pain catastrophizing in a fear-avoidance model of chronic pain. Pain. 2010 Jan; 148(1): 120-127. doi: 10.1016 / j. pain.2009.10.030. Epub 2009 Nov 26. PMID: 19944534.

[0166] 2. Quartana PJ, Campbell CM, Edwards RR. Pain catastrophizing: a critical review. Expert Rev Neurother. 2009 May;9(5):745-58. doi: 10.1586 / ern.09.34. PMID: 19402782; PMCID: PMC2696024

[0167] 3. A. Dey and G. Abowd, “Towards a better understanding of context and contextawareness”, Proceedings of the Workshop on the What, Who, Where, When and How of Context- Awareness, affiliated with the CHI 2000 Conf, on Human Factors in Computer Systems, New York, NY, 2000

[0168] REFERENCE NUMERALS

[0169] I Medical Device Communication System (MDCS)

[0170] 10 Implantable Medical Device (IMD)

[0171] I I IMD stimulator, e.g., SCS device with attached electrodes

[0172] 12 IMD wireless personal area network (WPAN) transceiver

[0173] 13 IMD sensor

[0174] 14 IMD electrical energy source

[0175] 16 IMD control circuit

[0176] 17 IMD memory

[0177] 18 IMD processor

[0178] 19 IMD electrode drive circuit

[0179] 20 charger for IMD

[0180] 24 charger energy source

[0181] 29 charger external mains power connection

[0182] 30 patient remote controller (patient remote)

[0183] 32 smartphone wireless personal area network (WPAN) transceiver

[0184] 35 smartphone cellular transceiver

[0185] 36 smartphone wireless local area network (WLAN) transceiver

[0186] 38 router

[0187] 40 telephone line / telephone network

[0188] 42 cellular network base station (cellular tower)

[0189] 50 internet connection

[0190] 60 remotely located data center (backend)

[0191] 70 portal

[0192] 80 lead arrangement

[0193] 82 leads

[0194] 84 electrodes

[0195] 100 computing apparatus

[0196] 105 communication bus 110 processor

[0197] 115 main memory

[0198] 120 secondary memory

[0199] 125 internal memory

[0200] 130 removable medium

[0201] 135 I / O interface

[0202] 140 communication interface

[0203] 145 external storage medium

[0204] 150 communication channel

[0205] 155 electrical communication signals

[0206] 160 baseband system

[0207] 165 radio system

[0208] 170 antenna system

[0209] 175 display

[0210] 200 pain

[0211] 201 sentiment

[0212] 202 cognitive state (cognition)

[0213] 203 affective state (affectivity)

[0214] 204 negative affectivity, e.g., pain catastrophizing

[0215] 205 positive affectivity, e.g., no fear of pain

[0216] 207 experience, e.g., pain experience

[0217] 208 pain score

[0218] 209 pain state

[0219] 210 improvement

[0220] 212 deterioration

[0221] 214 conversational data

[0222] 215 behavioral data

[0223] 216 transcription

[0224] 218 annotation

[0225] 220 decomposition

[0226] 222 large language model, LLM 224 natural language processing, NLP

[0227] 226 machine learning model, ML model

[0228] 228 inference and prediction

[0229] 230 diction assessment 232 dynamics model

[0230] 234 kinematic processing

Claims

Claims1. A computer-implemented method for evaluation of a patient suffering from pain, the method comprising: inputting patient contextual data (214, 215) collected from a patient suffering from pain; processing the contextual data in a machine learning model (226) trained to link patient contextual data of pain patients to at least one pain sentiment parameter to determine a probability value for the at least one pain sentiment parameter; and outputting at least one of an inference and a prediction (228) as a probability value of the at least one pain sentiment parameter, wherein the inference is a characteristic of the current state of the patient and the prediction is a characteristic of a predicted future state of the patient, wherein the patient contextual data includes conversational data collected from a patient interview.

2. The method of claim 1, wherein the pain sentiment parameters include parameters selected from at least one of the group: patient states, patient therapy efficacy, patient therapy outcomes, and patient therapy interventions.

3. The method of claim 1 or 2, wherein the at least one pain sentiment parameter output by the machine learning model is selected from the group: probability of therapy success, probability of the patient feeling optimistic, probability of the patient suffering from pain catastrophizing; and probability of the patient needing an intervention.

4. The method of any one of the preceding claims, wherein a large language model (222) and natural language processing (224) extract from the conversational data a quantitative value of at least one conversation-derived variable selected from the group: sentiment score, stress level, cognitive load, cognitive bias, affective bias,and mood, said at least one conversation-derived variable being supplied to the machine learning model as input.

5. The method of any one of the preceding claims, wherein the patient contextual data includes behavioral data collected from a device collecting data on patient behavior.

6. The method of claim 5, wherein a dynamics model (232) and kinematic processing (234) extract from the behavioral data a quantitative value of at least one behavioral variable selected from the group: posture, gait, sensory perception, and sensitivity, said at least one behavioral variable being supplied to the machine learning model as input.

7. The method of any one of the preceding claims, wherein the patient contextual data further includes a patient self-assessed pain score according to a numerical rating scale for pain.

8. The method of claim 7, wherein a value of a pain sentiment parameter relating to a current state of the patient is combined with the patient’s self-assessed pain score to classify the patient into one of a plurality of patient outcome groups, said groups including at least a first group for patients who are dissatisfied with their pain therapy and a second group for patients who are satisfied with their pain therapy.

9. The method of claim 8, wherein said groups further comprise at least a third group for patients who are neither satisfied nor dissatisfied with their pain therapy.

10. The method of claim 8 or 9, further comprising: providing a clinical recommendation for future treatment of the patient based on the probability value for the at least one pain sentiment parameter.

11. The method of claim 10, wherein the patient being evaluated is undergoing treatment with a spinal cord stimulation device (10) and wherein the clinical recommendationcomprises values for a set of control parameters for the spinal cord stimulation device.

12. The method of any one of the preceding claims, wherein based on a value of a pain sentiment parameter relating to a future state of the patient a clinical recommendation is output regarding future treatment.

13. The method of any one of the preceding claims, wherein the machine learning model incorporates a psychological model of pain that distinguishes between patients experiencing the sentiment of pain catastrophizing and those who are not.

14. A non-transitory computer readable medium (120, 125, 130, 145) storing instructions for performing a computer-implemented method according to any one of the preceding claims.

15. A computer apparatus (60; 100) having a memory (115) and at least one processor (110), the memory having loaded therein instructions for performing a computer- implemented method according to any one of the preceding claims, which, when executed by the at least one processor, cause the at least one processor to perform the computer-implemented method according to any one of the preceding claims.

Images