Method and Apparatus for Predicting Digital Therapeutic Response of Elderly Depression Patients Using Reaction Time Analysis Based on Ecological Momentary Assessment
By using response time data from EMA and preprocessing techniques, the method addresses subjective biases in EMA, enabling precise prediction of treatment responses and personalized strategies for elderly patients with depression.
Patent Information
- Application Number
- KR1020250140928
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2025-09-29
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2045-09-29
AI Technical Summary
Existing Ecological Momentary Assessment (EMA) methods for mental health management rely heavily on subjective self-reporting, leading to discrepancies and response bias, limiting the ability to quantitatively predict treatment effects and individual patient responsiveness, especially in elderly patients with depression.
A method and apparatus that utilize response time data from EMA to model a patient's response pattern, applying preprocessing techniques like outlier removal and within-person standardization, and fitting the data to an exponential decay curve to extract parameters for predicting digital treatment response, distinguishing between depressive and anxiety symptoms.
This approach provides objective indicators for evaluating treatment responsiveness, improves data quality by minimizing bias, and enables personalized treatment strategies by accurately predicting symptom relief and comorbidities in elderly patients with depression.
Smart Images

Figure 112025110852929-PAT00010_ABST
Abstract
Description
Technology Field
[0001] The present invention relates to the field of technology related to the mental health management of elderly patients with depression, and more specifically, to a method and apparatus for modeling a patient's response pattern by analyzing response time data collected from an Ecological Momentary Assessment (EMA) and predicting the response to a digital therapeutic intervention. The present invention can be utilized in the fields of digital therapeutic (DTx) applications, clinical mental health assessment, and artificial intelligence-based data analysis technology. Background Technology
[0002] In the field of mental health management, the Ecological Momentary Assessment (EMA) is widely used as a tool to assess a patient's daily state in real time. The EMA is conducted via a self-report method in which the patient responds to questions presented periodically or non-periodically through a user interface, such as a mobile device, thereby allowing the collection of the patient's emotional, cognitive, and behavioral states at different points in time.
[0003] However, because the existing EMA method relies heavily on respondents' subjective perceptions and reporting attitudes, there are issues such as discrepancies between the actual condition and the reported result even for the same patient, as well as response bias and fatigue from repeated measurements affecting the results. Therefore, relying solely on EMA has limitations in quantitatively predicting treatment effects or determining individual patient responsiveness to treatment.
[0004] Meanwhile, with the introduction of digital therapeutics (DTx) into the management of mental illness, it has been reported that patients can alleviate symptoms such as depression and anxiety by repeatedly using mobile applications. However, the magnitude and timing of these therapeutic responses vary among individual patients, and there is a problem in that there are limitations in terms of clinical application and resource allocation due to the lack of a methodology to distinguish between patient groups for whom treatment intervention is effective and those for whom it is ineffective. Prior art literature
[0005] Republic of Korea Published Patent Application No. 10-2023-0124786 (Date of publication: August 28, 2023) The problem to be solved
[0006] The present invention aims to solve the aforementioned problems. An embodiment according to the present disclosure aims to provide a technology capable of quantitatively evaluating and predicting the treatment responsiveness of elderly patients with depression by utilizing response time data collected during the process of performing an Ecological Momentary Assessment (EMA). Through this, the invention seeks to resolve the limitations of subjective self-reporting in the existing EMA and the problem of individual differences in digital therapeutic interventions (DTx), and to provide objective indicators that can distinguish treatment effects for each patient in advance.
[0007] In addition, one embodiment according to the present disclosure aims to provide a preprocessing technique comprising outlier removal based on an interquartile range rule and within-person standardization processing to remove outliers inherent in response time data and confounding effects arising from user interface (UI) bias.
[0008] An embodiment according to the present disclosure aims to provide a technology that models preprocessed response time data as an exponential decay curve over time, extracts parameters such as the rate of change in response time, the rate of adaptation, and the variance of response time, and utilizes these as input variables for predicting digital treatment response.
[0009] Furthermore, one embodiment according to the present disclosure aims to provide a technology capable of distinguishing between depression and anxiety and differentiating the presence of comorbidities by utilizing the differential characteristics in which a sharp decrease in the rate of change of reaction time and stabilization of variance are clearly observed in depressive symptoms, whereas no change is observed when the same indicator is applied in anxiety symptoms.
[0010] However, the problems to be solved in this disclosure are not limited to those mentioned above, and may be expanded in various ways without departing from the spirit and scope of this disclosure. means of solving the problem
[0011] One technical aspect of the present invention proposes a method for predicting the response to digital treatment in elderly patients with depression using response time analysis based on ecological momentary evaluation. The digital treatment response prediction method is performed by a digital treatment response prediction device and comprises: a step of collecting response time data for an ecological momentary evaluation performed by an elderly patient with depression through a user terminal; a step of generating a response pattern model of the elderly patient with depression by modeling the response pattern of the elderly patient with depression over time based on the response time data; and a step of predicting the response to digital treatment in the elderly patient with depression using the response pattern model.
[0012] In one embodiment, the ecological moment evaluation includes items presented periodically or non-periodically through the user terminal during the daily life of the elderly patient with depression and response data thereto, and the response data may include self-report evaluation data that records the emotional, cognitive, and behavioral states at each response point in real time.
[0013] In one embodiment, prior to the step of generating a response pattern model for the elderly patient with depression, the method may include: a step of removing outliers by applying a quartile rule to the response time data; and a step of performing standardization processing on the response time within an individual to control the confounding effect arising from bias based on the configuration of response positions per item.
[0014] In one embodiment, the reaction pattern model can be generated by fitting the change in reaction time over time to an exponential decay curve.
[0015] In one embodiment, the step of predicting the digital treatment response of the elderly patient with depression may include the step of extracting a plurality of parameters, including the rate of change of response time, the adaptation speed, or the variance of response time, from the exponential decay curve.
[0016] In one embodiment, the step of predicting the digital treatment response of the elderly patient with depression may include the step of determining the possibility of symptom relief of the elderly patient with depression as favorable when the rate of change in response time decreases rapidly above a threshold value.
[0017] In one embodiment, the step of predicting the digital treatment response of the elderly patient with depression may include: classifying the patient as a treatment responder when the adaptation speed is greater than or equal to a preset threshold value; and classifying the patient as a non-responder when the response time variance does not converge to or less than the threshold value.
[0018] In one embodiment, the step of predicting the response to digital treatment of the elderly patient with depression may include: a step of estimating the degree of improvement in depressive symptoms as good when the rate of change in response time decreases below a preset reduction threshold value and the variance in response time converges below a preset stability threshold value; and a step of differentially determining depressive symptoms and anxiety symptoms to determine whether there is a comorbidity between depression and anxiety when the rate of change in response time and the variance in response time are applied equally to an indicator related to anxiety symptoms, and the conditions for the reduction threshold value and the stability threshold value are not met in the indicator related to anxiety symptoms.
[0019] Another technical aspect of the present invention proposes a digital treatment response prediction device. The digital treatment response prediction device comprises at least one processor; and a memory for storing instructions. When the instructions are executed individually or collectively by the at least one processor, the processor performs the following operations: collecting response time data for an ecological moment assessment performed by an elderly patient with depression through a user terminal; modeling the response pattern of the elderly patient with depression over time based on the response time data to generate a response pattern model of the elderly patient with depression; and predicting the digital treatment response of the elderly patient with depression using the response pattern model. Effects of the invention
[0020] According to various embodiments of the present disclosure, by utilizing response time data collected during the EMA process, it is possible to provide a technology that can compensate for bias due to subjectivity in existing self-reported response data and objectively evaluate the treatment responsiveness of elderly patients with depression.
[0021] According to various embodiments of the present disclosure, by performing preprocessing steps such as outlier removal and intra-personal standardization on response time data, the data quality is improved and contamination effects resulting from the user interface (UI) configuration are eliminated, thereby providing the effect of increasing the reliability of treatment response analysis.
[0022] According to various embodiments of the present disclosure, by fitting the temporal variation of reaction time to an exponential decay curve model or the like to model the reaction pattern, the effect of deriving parameters that quantitatively characterize the cognitive processing efficiency and adaptation speed for each patient is provided.
[0023] According to various embodiments of the present disclosure, by utilizing extracted parameters as input variables in a digital treatment response prediction step, the potential for symptom relief in patients is quantitatively estimated, and the effect of clearly distinguishing between a treatment responder group and a non-responder group is provided.
[0024] According to various embodiments of the present disclosure, by applying the rate of change in reaction time and the dispersion index to depressive symptoms and anxiety symptoms, respectively, the difference in which an improvement pattern appears in depressive symptoms but not in anxiety symptoms is captured, thereby providing an effect that can differentiate the presence of comorbidities of depression and anxiety.
[0025] According to various embodiments of the present disclosure, by predicting a patient's responsiveness to treatment early, the intensity, frequency, and timing of digital therapeutic interventions can be optimized in advance, providing a substantial decision-making support effect that enables clinicians to establish patient-specific treatment strategies.
[0026] According to various embodiments of the present disclosure, the present invention provides the effect of simultaneously securing clinical utility and industrial utility value by precisely evaluating the therapeutic effect on elderly patients with depression in the field of digital therapeutics (DTx), reducing resource waste, and enabling personalized treatment for each patient.
[0027] The effects obtainable from the present disclosure are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art to which the present disclosure belongs from the description below. Brief explanation of the drawing
[0028] FIG. 1 is a block diagram illustrating the schematic configuration of a digital treatment response prediction device for elderly patients with depression according to one embodiment of the present invention. FIG. 2 is a flowchart illustrating a method for predicting the response to digital treatment in elderly patients with depression using an ecological moment evaluation-based response time analysis according to an embodiment of the present invention. FIG. 3 is a diagram showing the screen switching procedure of an ecological moment evaluation question presented through a user terminal (200) in a method for predicting the response to digital treatment of an elderly patient with depression using an ecological moment evaluation-based response time analysis according to one embodiment of the present invention. FIG. 4 is a flowchart illustrating the procedure for removing outliers and standardizing reaction time data in a method for predicting digital treatment response in elderly patients with depression using reaction time analysis based on ecological moment evaluation according to an embodiment of the present invention. FIG. 5 is an example histogram illustrating the process of removing outliers from reaction time data using the interquartile rule (IQR rule) according to an embodiment of the present invention. FIG. 6 is a flowchart illustrating a procedure for generating a curve-fitting-based response pattern model according to an embodiment of the present invention. Figure 7 is a graph showing the results of modeling by applying an exponential decay curve to individual reaction time data according to an embodiment of the present invention. FIG. 8 is a flowchart illustrating the process of classifying treatment response groups based on curve parameters and calculating prediction results according to one embodiment of the present invention. FIG. 9 is a flowchart for explaining the process of differentiating between depressive symptoms and anxiety symptoms using reaction time change rate and dispersion index according to one embodiment of the invention, and further determining whether there is a comorbid depression / anxiety disorder. FIG. 10 is a diagram illustrating the process of a digital treatment response prediction device (100) according to one embodiment of the present invention classifying a responder group and a non-responder group by applying a Minimal Detectable Change (MDC) threshold and comparing and analyzing the EMA response time curves of each group. FIG. 11 is a series of graphs illustrating the process of analyzing the gradual difference appearing in the EMA (Experience Sampling Method) reaction time curve according to the magnitude of the intervention effect according to one embodiment of the present invention. Specific details for implementing the invention
[0029] Hereinafter, embodiments of the present disclosure are described in detail with reference to the drawings so that those skilled in the art can easily practice them. However, the present disclosure may be embodied in various different forms and is not limited to the embodiments described herein. In relation to the description of the drawings, the same or similar reference numerals may be used for identical or similar components. Furthermore, in the drawings and related descriptions, descriptions of well-known functions and configurations may be omitted for clarity and brevity.
[0030] The various embodiments of this document and the terms used therein are not intended to limit the technical features described in this document to specific embodiments, and should be understood to include various modifications, equivalents, or substitutions of said embodiments. In connection with the description of the drawings, similar reference numerals may be used for similar or related components. The singular form of a noun corresponding to an item may include one or more of said items unless the relevant context clearly indicates otherwise. In this document, phrases such as "A or B," "at least one of A and B," "at least one of A or B," "A, B or C," "at least one of A, B and C," and "at least one of A, B, or C" may each include any one of the items listed together in the corresponding phrase, or all possible combinations thereof. Terms such as "first," "second," or "first" or "second" may be used simply to distinguish said components from other said components and do not limit said components in any other aspect (e.g., importance or order). Where any (e.g., 1st) component is referred to as "coupled" or "connected" to another (e.g., 2nd) component, with or without the terms "functionally" or "communicationly," it means that said any component may be connected to said other component directly (e.g., via a wire), wirelessly, or through a third component.
[0031] The term “module” as used in the various embodiments of this document may include a unit implemented in hardware, software, or firmware, and may be used interchangeably with terms such as logic, logic block, component, or circuit, for example. A module may be a component formed integrally, or a minimum unit of said component or a part thereof that performs one or more functions. For example, according to one embodiment, a module may be implemented in the form of an application-specific integrated circuit (ASIC).
[0032] Various embodiments of this document may be implemented as software (e.g., a program) comprising one or more instructions stored in a storage medium (e.g., memory) readable by a machine or device. For example, the processor of the machine or device may call at least one of the one or more instructions stored from the storage medium and execute it. This enables the machine to operate to perform at least one function according to the at least one called instruction. The one or more instructions may include code generated by a compiler or code that can be executed by an interpreter. The storage medium readable by a machine may be provided in the form of a non-transitory storage medium. Here, "non-transitory" simply means that the storage medium is a tangible device and does not contain a signal (e.g., electromagnetic waves), and this term does not distinguish between cases where data is stored semi-permanently and cases where it is stored temporarily in the storage medium.
[0033] According to one embodiment, the method according to the various embodiments disclosed herein may be provided by being included in a computer program product. The computer program product may be traded between a seller and a buyer as a product. The computer program product may be distributed in the form of a device-readable storage medium (e.g., compact disc read-only memory (CD-ROM)), or distributed online (e.g., download or upload) through an application store or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product may be temporarily stored or temporarily created on a device-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.
[0034] According to various embodiments, each component (e.g., module or program) of the components described above may include a singular or multiple entities, and some of the multiple entities may be separated and placed in other components. According to various embodiments, one or more of the components or operations of the aforementioned components may be omitted, or one or more other components or operations may be added. Generally or additionally, multiple components (e.g., module or program) may be integrated into a single component. In this case, the integrated component may perform one or more functions of each of the multiple components in the same or similar manner as those performed by the corresponding component among the multiple components prior to integration. According to various embodiments, operations performed by the module, program, or other components may be executed sequentially, in parallel, iteratively, or heuristically, or one or more of the operations may be executed in a different order, omitted, or one or more other operations may be added.
[0035] In this disclosure, the term "processor" may refer to hardware capable of performing functions and operations according to each designation described herein, computer program code capable of performing specific functions and operations, or an electronic recording medium loaded with computer program code capable of performing specific functions and operations. According to an embodiment, the operation of the processor may be defined and / or interpreted as the operation of a knowledge graph adjustment device, but is not limited thereto. The processor may refer to a functional and / or structural combination of hardware for carrying out the technical concept of this disclosure and / or software for driving said hardware.
[0037] The embodiments of the present invention described below are designed to be suitable for predicting digital treatment responses, particularly for elderly patients with depression, but are not limited thereto. Compared to young adults, the elderly have characteristics such as slower response speeds, greater variability in response, and a tendency for bias to occur during the use of user interfaces. Accordingly, it is difficult to accurately predict the digital treatment response of elderly patients using only simple survey score-based evaluation methods.
[0038] Embodiments of the present invention utilize the reaction time data itself as a key indicator to reflect the characteristics of elderly patients, and by applying techniques such as the Interquartile Quartile (IQR) rule, Robust Scaling, and Winsorizing to remove outliers and standardize the data, distortions caused by slow or irregular responses commonly observed in the elderly can be minimized. In addition, input biases prominent in the elderly population, such as hand tremors or button position dependency, are controlled through a user interface position bias removal technique.
[0039] Furthermore, the embodiments of the present invention utilize an exponential curve-fitting-based response pattern model rather than a simple linear trend, thereby accurately reflecting the non-linear and delayed treatment adaptation patterns observed in elderly patients. Moreover, by eliminating potential confounding effects caused by emotional intensity, clarity, or interface configuration through intra-individual response time standardization (Z-score transformation), treatment responses can be stably compared and analyzed even in groups of elderly patients with significant individual differences.
[0040] In addition, by extracting parameters such as the rate of change, the speed of adaptation, and the variance of response from curve fitting, it is possible to quantitatively distinguish between depressive and anxiety symptoms in elderly patients. For example, elderly patients with depressive symptoms may exhibit a gradual and delayed response pattern and high variance, whereas elderly patients with anxiety symptoms may exhibit a rapid initial response and a short-term overreaction, and the present invention enables more precise patient-specific prediction by reflecting these characteristics.
[0042] FIG. 1 is a block diagram illustrating the schematic configuration of a digital treatment response prediction device for elderly patients with depression according to one embodiment of the present invention.
[0043] As illustrated in FIG. 1, the digital treatment response prediction device for elderly patients with depression according to the present invention (hereinafter abbreviated as ‘digital treatment response prediction device’) (100) may include one or more processors (101) and memory (102).
[0044] The digital treatment response prediction device (100) can be connected to communicate with a user terminal (200).
[0045] The user terminal (200) can be implemented as a smartphone, tablet, wearable device, etc., and performs the function of periodically or non-periodically presenting Ecological Momentary Assessment (EMA) questions to the patient and collecting the patient's self-reported responses and corresponding response time data.
[0046] For example, response data may consist of data that reflects the patient's emotional state, cognitive processing, and behavioral responses in real time.
[0047] A digital treatment response prediction device (100) receives response time data transmitted from a user terminal (200) and can perform data preprocessing, response pattern modeling, parameter extraction, or treatment response prediction thereon.
[0048] For example, a digital treatment response prediction device (100) can control the mixing effect according to the order or position of item presentation by applying an interquartile rule (IQR rule) to response time data to remove outliers and standardizing response times within an individual (see claim 3). Through this, data quality can be ensured and the reliability of the analysis can be improved.
[0049] For example, a digital treatment response prediction device (100) can generate a patient response pattern model by fitting the change in the patient's response time over time to an exponential decay curve. This response pattern model can mathematically describe the process of the patient's response time gradually shortening or stabilizing, and can be used as basic data to evaluate the speed of adaptation and the possibility of symptom relief.
[0050] For example, a digital treatment response prediction device (100) can extract parameters such as the rate of change in response time, the rate of adaptation, and the variance of response time from a response pattern model, and can quantitatively predict the digital treatment response of a specific patient based on this. For example, if the rate of change in response time decreases rapidly, the possibility of symptom relief can be evaluated as high, and if the rate of adaptation is above a threshold value, the patient can be classified as a treatment responder group, and if the variance of response time does not converge, the patient can be classified as a non-responder group.
[0051] For example, a digital treatment response prediction device (100) can differentiate between depressive symptoms and anxiety symptoms and determine whether there is a comorbidity by comparing the extracted response time indicator with an anxiety symptom indicator, and if improvement in depressive symptoms is statistically significant but the same improvement does not appear in anxiety symptoms. This allows a clinician to evaluate the response to treatment for depression while simultaneously performing a precise pathological distinction from anxiety.
[0052] For example, the digital treatment response prediction device (100) may store a long-term tracking database of patients, through which long-term treatment trends of patients can be modeled or comparative analysis between individual patient groups can be performed. For instance, by accumulating patients' weekly or monthly EMA records, not only short-term response patterns but also medium- and long-term symptom change curves can be estimated.
[0053] For example, a digital treatment response prediction device (100) is connected to a cloud-based server to integrate and analyze data from a large patient group and apply machine learning techniques to continuously improve the accuracy of predicting individual response patterns. This allows for the establishment of customized treatment strategies for each patient, as well as the verification of the effectiveness of digital therapeutics (DTx) and the analysis of clinical trial data.
[0054] The digital treatment response prediction device (100) may include a processor (101) and a memory (102).
[0055] The processor (101) operates operatively with the memory (102) to perform various processing logics of the digital treatment response prediction device (100). For example, the processor (101) can receive response time data transmitted from the user terminal (200) and execute a program stored in the memory (102) to perform data preprocessing, response pattern modeling, parameter extraction, or treatment response prediction.
[0056] Such a processor (101) may include, as an example, at least one of a microprocessor, a central processing unit, a processor core, a multi-core processor, a multiprocessor, an ASIC (Application-Specific Integrated Circuit), or an FPGA (Field Programmable Gate Array), but is not limited thereto.
[0057] The memory (102) can store data required for various processing steps performed by the processor (101). The memory (102) can store instructions (or programs) that can be executed by the processor (110).
[0058] The memory (120) may include volatile memory or non-volatile memory. Volatile memory may be implemented as DRAM (dynamic random access memory), SRAM (static random access memory), T-RAM (thyristor RAM), Z-RAM (zero capacitor RAM), or TTRAM (Twin Transistor RAM). Non-volatile memory can be implemented as EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, MRAM (Magnetic RAM), Spin-Transfer Torque (STT)-MRAM, Conductive Bridging RAM (CBRAM), FeRAM (Ferroelectric RAM), PRAM (Phase change RAM), Resistive RAM (RRAM), Nanotube RRAM, Polymer RAM (PoRAM), Nano Floating Gate Memory (NFGM), holographic memory, Molecular Electronic Memory Device, or Insulator Resistance Change Memory.
[0059] In this way, the digital treatment response prediction device (100) moves away from simple EMA self-report data-based analysis and precisely utilizes objective response time indicators, thereby enabling effective evaluation and prediction of the treatment response of elderly patients with depression.
[0061] FIG. 2 is a flowchart illustrating a method for predicting the response to digital treatment in elderly patients with depression using an ecological moment evaluation-based response time analysis according to an embodiment of the present invention.
[0062] The method of Fig. 2 is performed in the digital treatment response prediction device (100) shown in Fig. 1.
[0063] First, in step S210, the digital treatment response prediction device (100) can collect response time data for an ecological moment evaluation performed by an elderly patient with depression during daily life through a user terminal (200).
[0064] Here, the ecological moment assessment includes items and response data presented periodically or non-periodically through a user terminal within the patient's daily context, and includes self-report assessment data that records the emotional, cognitive, and behavioral states at each response point in real time.
[0065] The reaction time data collected in this way serves as source data that can objectively reflect the emotional and cognitive responses occurring in the patient's actual living environment.
[0066] Next, in step S220, the digital treatment response prediction device (100) can perform a preprocessing process on the collected response time data.
[0067] For example, the preprocessing process includes a step of removing outliers from the reaction time data by applying the interquartile rule (IQR rule), and a step of standardizing the reaction time within an individual to control for confounding effects resulting from biases caused by the presentation order of items or the arrangement of the interface.
[0068] Through such preprocessing, data distortion can be minimized, and the unique response characteristics of the patient can be stably reflected.
[0069] Next, in step S230, the digital treatment response prediction device (100) generates a response pattern model by modeling the patient's response pattern over time based on the preprocessed response time data.
[0070] In one embodiment, the response pattern model can be generated by fitting the change in response time with an exponential decay curve, thereby allowing key characteristics such as the patient's adaptation speed, the rate of change in response time, and the variance of response time to be extracted as parameters.
[0071] In step S240, the digital treatment response prediction device (100) predicts the digital treatment response of an elderly patient with depression using a generated response pattern model.
[0072] In one embodiment, the digital treatment response prediction device (100) may determine that the possibility of symptom relief for the patient is good when the rate of change in response time decreases rapidly above a threshold value.
[0073] In one embodiment, the digital treatment response prediction device (100) may classify a patient as a treatment responder if the adaptation speed is greater than or equal to a reference value, and as a non-responder if the response time variance does not converge to below a reference value.
[0074] In one embodiment, the digital treatment response prediction device (100) analyzes the rate of change in response time and the variance in response time in parallel with an indicator related to anxiety symptoms, and if an improvement pattern appears in depressive symptoms but no significant improvement appears in anxiety symptoms, it can differentiate between depression and anxiety and determine whether there is a comorbidity. By doing so, it can provide precise diagnostic value that goes beyond simply predicting whether depressive symptoms improve and determines clinical differences with accompanying anxiety symptoms.
[0075] In addition, in one embodiment, the digital treatment response prediction device (100) can utilize parameters derived from a response pattern model (e.g., rate of change, rate of adaptation, variance) as input variables to quantitatively score the possibility of symptom relief or automatically classify patients into a treatment responder group and a non-responder group. Such classification results can be used by a clinician to set the timing, intensity, and direction of treatment for patient-specific digital treatment intervention.
[0077] FIG. 3 is a diagram showing the screen switching procedure of an ecological moment evaluation question presented through a user terminal (200) in a method for predicting the response to digital treatment of an elderly patient with depression using an ecological moment evaluation-based response time analysis according to one embodiment of the present invention.
[0078] As shown in the left screen of FIG. 3, the user terminal (200) presents the patient with an emotion-related self-report question such as "How have you been in the last 24 hours?"
[0079] The patient performs a response by touching one of several options, and for example, the drawing illustrates the case where the middle value of a 5-point scale is selected. During this process, the time at which the patient selects the response and the time at which the touch input is completed are recorded, and the interval between the two times is collected as response time data.
[0080] Afterwards, when the patient clicks the "Next" button, the screen switches as shown on the right screen of Fig. 3, and this time, behavioral self-report questions such as "How was your appetite over the past 24 hours?" are presented.
[0081] In this way, the user terminal (200) periodically or non-periodically presents various self-report questions that can reflect the emotional, cognitive, and behavioral states of the patient during daily life, and automatically collects response time data during the response process for each question.
[0082] In one embodiment, the user terminal (200) can record the response time in milliseconds (ms) from immediately after the presentation of the question until the completion of the response selection, thereby eliminating system errors caused by network latency or UI response speed to reflect only the patient's unique response characteristics. Additionally, to minimize the order effect or learning effect among multiple questions, a method of randomly shuffling the presentation order of the questions or counterbalancing within a specific range may be applied.
[0083] In one embodiment, the user terminal (200) may not only record the response button click time, but may also log detailed input patterns such as the patient's finger movement trajectory immediately after screen transition, points of hesitation occurring in the middle, and whether multiple clicks occurred, and these detailed input patterns may be used as auxiliary indicators in the interpretation of response time data. For example, even if the same response time is observed, if frequent corrections or hesitation are observed during the input process, it may be interpreted as a sign of cognitive load or emotional instability.
[0084] The digital treatment response prediction device (100) of the present invention collects each question presented on a user terminal (200) as shown in FIG. 3 and response time data for the same, and by linking this data to preprocessing, modeling, and prediction steps, it can quantitatively estimate whether the patient responds to treatment and the possibility of symptom improvement.
[0086] FIG. 4 is a flowchart illustrating the procedure for removing outliers and standardizing reaction time data in a method for predicting digital treatment response in elderly patients with depression using reaction time analysis based on ecological moment evaluation according to an embodiment of the present invention.
[0087] First, in step S410, reaction time data measured during the patient's response to the ecological moment evaluation questions presented through the user terminal (200) is collected.
[0088] This reaction time data includes the time interval from the presentation of each item to the completion of the response, and is recorded as individual data points for emotional, cognitive, and behavioral self-report items. For example, the reaction time data includes the reaction speed when selecting a response measured on a Likert scale for multiple items such as mood, appetite, sleep, and overall evaluation.
[0089] Next, in step S420, the digital treatment response prediction device (100) applies an Interquartile Range (IQR) rule to remove outliers from the response time data.
[0090] At this time, outlier removal is performed by calculating a lower limit (Q1 - 0.5ХIQR) and an upper limit (Q3 + 1.5ХIQR) by considering the skewness of the distribution for each item, and identifying values outside the range as outliers and removing them.
[0091] For example, the first quartile (Q1) and the third quartile (Q3) of the reaction time data can be calculated, and after calculating IQR = Q3 - Q1, values that are smaller than Q1 - 1.5 ХIQR or larger than Q3 + 1.5 ХIQR can be determined as outliers and removed.
[0092] This method excludes extreme reaction times that may occur due to individual patient response characteristics (e.g., unintended screen touches, prolonged delays due to loss of concentration), thereby ensuring that only reliable data is obtained for analysis.
[0093] In step S430, the digital treatment response prediction device (100) performs individual standardization processing on the refined response time data.
[0094] For example, reaction times within the same individual can be normalized based on the within-individual distribution by converting them into mean and standard deviation units (Z-transform) for each response level (e.g., a 1–7 point Likert scale). For example, the standardized reaction time Z for attempts k performed by individual i at response level j. ij (k) is defined as in [Mathematical Formula 1] below.
[0095] [Mathematical Formula 1]
[0096]
[0097] Here, standardized reaction time Z ij (k) is the raw data reaction time recorded by individual i at response level j at trial k, and is the average response time at the same response level j, σ ij is the standard deviation of the reaction time under the corresponding conditions.
[0098] This allows for controlling for differences in reaction times based on individual emotional intensity or clarity, and eliminates biases that may arise from the user interface (UI), such as faster reaction times due to proximity to the "Next" button when selecting extreme scores at either end of a scale. Therefore, by correcting for differences in reaction speed and variance between individuals, more reliable subsequent analysis becomes possible.
[0099] In step S440, outliers are removed and standardized refined reaction time data is output through the process described above. This refined data is used as input to generate a patient's reaction pattern model in a subsequent step (S230, see FIG. 2).
[0100] In one embodiment, the refined response time data is not stored merely as a time value, but can be stored in a structured form along with metadata such as the type of response item (emotional, cognitive, or behavioral), time of day (morning, evening, etc.), and repetition number. Through this, even the same response time value can be interpreted in a multi-layered manner by considering the context of the item or the temporal background.
[0101] In one embodiment, to further enhance the stability of response time data, a multi-stage normalization method is applied that combines Robust Scaling and Winsorizing techniques with IQR rule-based outlier removal, thereby maximizing the reliability and interpretability of the data. For example, by cutting off the top 1% and bottom 1% of extreme values or replacing them with boundary values, distortion of the data distribution can be minimized and the stability of the modeling can be ensured.
[0102] Specifically, in this embodiment, outlier removal based on the Interquartile Range (IQR) rule is performed in step S420. For example, the first quartile (Q1) and the third quartile (Q3) are calculated for a specific patient's reaction time data set, and then IQR = Q3 - Q1 is calculated. Then, values smaller than Q1 - 1.5 χ IQR or values larger than Q3 + 1.5 χ IQR are considered outliers and removed. This process can effectively exclude statistically extreme delayed reactions or abnormally fast responses (e.g., accidentally touching the screen).
[0103] Subsequently, before entering step S430, a robust scaling technique is additionally applied in this embodiment. Robust scaling is a method of transforming data based on the median and IQR rather than the mean and standard deviation, which is not sensitive to extreme values and can stably maintain the relative distribution of the data. For example, a transformation as shown in the following [Equation 2] is performed for reaction time x.
[0104] [Mathematical Formula 2]
[0105]
[0106] Here, Median(X) is the median response time of the patient, and IQR(X) is the interquartile range. Since this method guarantees scaling independent of patient-specific characteristics (such as cases where the average response time is very long or short), the response pattern model can ensure generalization performance across various patient groups without being biased toward a specific patient population.
[0107] Additionally, in this embodiment, the Winsorizing technique may be applied in parallel. Winsorizing is a method that replaces extreme values with predefined percentile boundary values (e.g., top 1%, bottom 1%) rather than simply removing them. For instance, values corresponding to the top 1% of reaction time data are replaced with the 99th percentile value, and values corresponding to the bottom 1% are replaced with the 1st percentile value. This approach allows for the preservation of clinically meaningful patterns by mitigating distribution distortion caused by extreme values while minimizing data loss.
[0108] In this embodiment, as a result of sequentially applying the IQR rule, robust scaling, and Winsorizing, the skewness and kurtosis of the data are statistically stabilized, and the goodness of fit of the exponential decay curve fitting used in the response pattern model is significantly improved. In particular, reliable curve parameters (rate of change, adaptation speed, variance, etc.) can be calculated even in response time data of elderly patients with many extreme values, which has the effect of significantly improving the accuracy of predicting treatment response in subsequent steps.
[0109] In addition, the present embodiment may be configured so that a multi-stage normalization technique is performed in parallel within an automated preprocessing module. For example, the digital treatment response prediction device (100) may implement an ensemble preprocessing technique that executes outlier removal, robust scaling, and Winsorizing as independent subroutines, respectively, and then dynamically selects an optimal refined data set. This provides an intelligent data processing environment that automatically applies an optimal normalization strategy suitable for the data characteristics of a specific patient group.
[0111] FIG. 5 is an example histogram illustrating the process of removing outliers from reaction time data using the interquartile rule (IQR rule) according to an embodiment of the present invention.
[0112] As illustrated in FIG. 5, in one embodiment of the present invention, outlier detection is performed based on response time data collected from an ecological moment assessment (EMA) in order to accurately predict the digital treatment response of elderly patients with depression.
[0113] Outlier detection is performed based on the quartiles of the entire dataset. Specifically, the first quartile (Q1) and third quartile (Q3) of the reaction time data are calculated, and the quartile range (IQR = Q3 - Q1) is derived from them. Subsequently, by defining the lower limit as Q1 - (0.5 Х IQR) and the upper limit as Q3 + (1.5 Х IQR), reaction time values below the lower limit or above the upper limit are identified as outliers. Unlike the traditional Tukey rule (Q1 - 1.5 Х IQR, Q3 + 1.5 Х IQR), this reflects the specificity of fast reaction times observed in the elderly population by applying a relaxed coefficient of 0.5 to the lower limit.
[0114] In addition, in one embodiment of the present invention, considering that the distribution of reaction time data is skewness, the IQR rule can be applied differentially to each data. For example, outlier boundaries are set for feeling reaction time data in the range of 2.11 seconds to 22.45 seconds, appetite reaction time data in the range of 1.8 seconds to 16.8 seconds, feelingOfSleep reaction time data in the range of 1.7 seconds to 19.7 seconds, and generallyRT data in the range of 6.2 seconds to 64.4 seconds.
[0115] Each histogram shown in Fig. 5 visually represents the results of data collection for each patient, and the intervals are divided based on the median, quartiles (Q1, Q3), and percentiles (1%, 5%, 95%, 99%). The intervals represented by colored areas are configured to allow for an intuitive verification of the application range of outlier detection rules.
[0116] According to one embodiment of the present invention, by removing outliers from response time data, noise data caused by UI usage habits or temporary decreases in concentration can be filtered out, and the actual cognitive response characteristics of the patient can be reflected more stably. Furthermore, this refined data is subsequently utilized in the training of a response pattern model to improve the accuracy of predicting digital treatment responses in elderly patients with depression.
[0117] As described above, in one embodiment, robust scaling or Winsorizing techniques may be applied in parallel to the data remaining after outlier removal. This allows for precise reflection of minute changes in treatment response patterns while minimizing distortion of response time data. For example, if extreme delay values of 60 seconds or more in the overall evaluation data are replaced with 95th percentile values through Winsorizing, the data distribution is stabilized, and the predictive performance of the model can be significantly improved.
[0119] FIG. 6 is a flowchart illustrating a procedure for generating a curve-fitting-based response pattern model according to an embodiment of the present invention.
[0120] One embodiment of FIG. 6 relates to one embodiment of modeling through exponential decay curve fitting.
[0121] Referring to FIG. 6, the digital treatment response prediction device (100) checks the response time data collected through the patient's user terminal (200) in step S610 and refined through the process of removing outliers and standardization.
[0122] In one embodiment, the digital treatment response prediction device (100) may apply Winsorizing to mitigate extreme values of the reaction time distribution along with outlier removal according to IQR rules, and may also apply robust scaling based on the median and interquartile range. By applying these multiple preprocessing techniques, it is possible to respond robustly to extreme delay values or input errors that frequently occur in the reaction time distribution of the elderly, and to increase the stability and prediction accuracy of the subsequent curve fitting step.
[0123] In step S620, the digital treatment response prediction device (100) performs exponential decay curve fitting on the response time data.
[0124] Specifically, reflecting the characteristic that the change in reaction time decreases non-linearly over time, the reaction time RT(t) can be modeled in the form of an exponential function as shown in [Equation 3] below.
[0125] [Mathematical Formula 3]
[0126]
[0127] Here, α is the initial response lag level, β is the adaptation rate over time, and γ is a constant representing the long-term steady state.
[0128] In step S630, the digital treatment response prediction device (100) extracts curve parameters (e.g., rate of change, adaptation rate, variance, etc.) from the curve fitting result.
[0129] Here, the rate of change indicates how quickly a patient's treatment response improves, and the variance signifies the consistency of the patient's response time data. These parameters are utilized as indicators to quantitatively express the characteristics of treatment response for each patient.
[0130] In step S640, the digital treatment response prediction device (100) stores the extracted parameters in the patient's response pattern model. This response pattern model can be used to predict the patient's future digital treatment response and to estimate long-term treatment effects.
[0131] According to one embodiment, a Bayesian curve fitting technique can be applied during the curve parameter extraction process to reflect the prior distribution for each patient. This enables stable and highly reliable curve estimation even for patients with limited data. Furthermore, for patients with high variability in treatment response, the Markov Chain Monte Carlo (MCMC) technique can be utilized to estimate the lagging distribution of parameters, thereby presenting a confidence interval for predicting the response of individual patients.
[0133] Figure 7 is a graph showing the results of modeling by applying an exponential decay curve to individual reaction time data according to an embodiment of the present invention.
[0134] As shown in Fig. 7, after the reaction time data of each subject is standardized, the change in reaction time over time (or the number of repeated trials) can be modeled through exponential curve fitting.
[0135] That is, the digital treatment response prediction device (100) can generate a response pattern model by fitting response time data, refined through an outlier removal and standardization process, to an exponential decay curve. Subsequently, by extracting parameters θa (initial response level), θb (adaptation speed), and θc (stability value) from this response pattern model, the characteristics of the patient's symptom change can be quantitatively calculated. Through this linked processing, the curve fitting step and the parameter extraction step can be implemented as a consistent pipeline.
[0136] The exponential decay curve used in this embodiment can be defined as shown in [Equation 4] below.
[0137] [Mathematical Formula 4]
[0138]
[0139] Here, θa represents the magnitude of the initial reaction time (latency) and signifies the difference in reaction time after practice compared to before practice. θb is an adaptation rate parameter indicating how quickly reaction time improves with each iteration. θc represents the stable reaction time reached after a sufficient number of iterations. The R² value indicates the goodness of fit of the curve fitting and evaluates how accurately the model explains the actual reaction time data.
[0140] In particular, elderly patients have the characteristic of having longer response times and larger standard deviations compared to general adults. The digital treatment response prediction device (100) takes into account this response variability of the elderly and utilizes the stability value (θc) of the exponential decay curve and the convergence of the response time variance as key indicators, rather than a simple average value. Through this, it is possible to make a prediction that reflects the characteristics of cognitive decline in the elderly and the time-dependent fatigue effect.
[0141] Graph (a) of Figure 7 (ID A) shows a pattern in which the initial response time is long and then decreases rapidly. This reflects the characteristic of the elderly showing a delayed initial response during the learning effect or digital therapy adaptation process, but stabilizing quickly through repeated trials. On the other hand, graph (b) (ID B) shows a pattern in which the initial response time is relatively short but decreases gradually, stabilizing slowly over a long period. This is an example showing that the speed of treatment adaptation and stable response patterns vary by individual.
[0142] As such, the present invention can quantitatively extract three key parameters—difference in initial response time, adaptation speed, and final stable response level—through individual response pattern modeling, thereby enabling precise prediction of digital treatment responses in the elderly and objective evaluation of differences based on disease characteristics such as depression and anxiety. In particular, the Levenberg-Marquardt least squares optimization algorithm is used during the curve fitting process to ensure stable convergence, and parameter bounds are adjusted according to the characteristics of each data distribution to prevent overfitting.
[0143] Therefore, response pattern modeling through exponential curve fitting can significantly improve the accuracy of digital therapy prediction for the elderly, as it can capture individual treatment response trends more precisely than indicators based on simple averages or medians.
[0145] FIG. 8 is a flowchart illustrating the process of classifying treatment response groups based on curve parameters and calculating prediction results according to one embodiment of the present invention.
[0146] Referring to FIG. 8, in step S810, the digital treatment response prediction device (100) evaluates whether the subject shows the possibility of symptom relief based on the rate of change (θa, θb, θc) of the individual response time curve calculated through curve fitting.
[0147] At this time, the rate of change reflects the degree of reduction and stabilization in subsequent trials compared to the initial response time, which has been verified as an EMA-based response speed indicator in the elderly depressed patient group.
[0148] Next, in step S820, the digital treatment response prediction device (100) classifies the subjects into a treatment responder group (Responder) and a non-responder group (Non-Responder) based on the adaptation rate (θb) and variance (σ²) of the response rate.
[0149] For example, in this process, the Minimal Detectable Change (MDC) criterion can be used to identify statistically significant changes between groups, rather than just simple average differences.
[0150] [Table 2]
[0151]
[0152] [Table 3]
[0153]
[0154] For example, as shown in [Table 2] above, according to the meta-analysis results based on Gonzalez et al. (2024), the MDC95 value calculated from the entire GDS-15 (Geriatric Depression Scale) dataset was reported to be 3.81 (95% CI: 3.59-4.04), which represents the minimum amount of change determined to be clinically valid beyond measurement error. Therefore, the digital treatment response prediction device (100) can classify the subject as a treatment responder if the amount of change extracted from the subject's EMA response time curve exceeds this threshold. Conversely, if the amount of change falls below the threshold, it is classified as a non-responder.
[0155] In one embodiment, confounding variables such as age (Mean age), sex (Sex), proportion with major depression classification, and hospitalization status (Setting) can be controlled by applying the meta-regression analysis results presented in [Table 3]. In particular, the proportion with major depression classification showed statistical significance of 0.29 (95% CI: 0.17-0.41, p<0.05), so the digital treatment response prediction device (100) can improve response prediction accuracy by interacting with the patient's initial severity.
[0156] [Table 4]
[0157]
[0158] Finally, in step S830, the digital treatment response prediction device (100) synthesizes the previously classified results to produce a final prediction result. For example, in [Table 4], a change in response time of the average Adjusted MDC = -5.5 level was observed for the Respondent group (n=14), whereas for the Non-Respondent group (n=45), the change was minimal or remained at a level that was not statistically significant.
[0159] In one embodiment, the digital treatment response prediction device (100) may be configured to automatically recalibrate the MDC threshold while updating the patient's EMA response time curve in real time. For example, a preliminary MDC may be estimated using only data from the first 10 days, and then, as data from the subsequent 20 days accumulates, the confidence interval of the MDC95 value may be narrowed, and the accuracy of classifying responders / non-responders may be gradually improved. This enables early intervention and allows for a preemptive response, particularly in the elderly patient group, without missing a treatment response.
[0160] In one embodiment, the prediction results can be extended beyond simple binary classification (responder / non-responder) to more granular clusters, such as partial responders and delayed responders. For example, elderly individuals who show a slow initial response but rapid stabilization after a certain point in time can be classified as a 'delayed responder,' which is useful for establishing clinical intervention strategies. Additionally, prediction accuracy can be further improved by applying machine learning-based classification models (e.g., logistic regression, random forest, neural network models) and utilizing multidimensional parameters such as rate of change, adaptation speed, variance, and stable response time as inputs.
[0161] Therefore, the process according to Fig. 8 is a key procedure for converting the results of individual response pattern modeling and curve parameter extraction of the present invention into clinically interpretable predictive indicators, and is a characteristic element that enables early determination of whether the elderly are adapting to digital treatment and providing customized treatment strategies.
[0163] FIG. 9 is a flowchart for explaining the process of differentiating between depressive symptoms and anxiety symptoms using reaction time change rate and dispersion index according to one embodiment of the invention, and further determining whether there is a comorbid depression / anxiety disorder.
[0164] Referring to FIG. 9, in step S910, the digital treatment response prediction device (100) applies the rate of change and variance indicators of response time to evaluation indicators related to depressive symptoms. At this time, if the rate of change (θb) and variance values calculated during the exponential decay curve fitting and parameter extraction process of the subject's response time meet or exceed a pre-set threshold for depressive symptoms, it can be estimated that the subject shows potential for improvement in depressive symptoms.
[0165] Next, in step S920, the digital treatment response prediction device (100) applies the same reaction time change rate and variance indicators to the anxiety symptom-related indicators. If the criteria for depressive symptoms are not met but the criteria for anxiety symptoms are met, the subject is determined to exhibit anxiety symptoms. By cross-applying the same reaction time-based parameters to both depression and anxiety symptom groups in this manner, the present invention can more precisely capture anxiety symptoms that may be overlooked in conventional survey-based evaluations.
[0166] Finally, in step S930, the digital treatment response prediction device (100) combines the judgment results from the preceding two steps to determine whether there is a comorbid depression / anxiety disorder. For example, if the response time indicator meets both the depression and anxiety criteria, it may be classified as a comorbid disorder group, which is used as important basic data when establishing a clinical treatment strategy.
[0167] Therefore, the procedure illustrated in Fig. 9 does not stop at simply evaluating whether depressive symptoms improve, but extends to differential differentiation between depression and anxiety and detection of comorbidities, thereby enabling more precise and personalized digital treatment prediction for complex emotional disorders commonly found in elderly subjects.
[0168] This effect is further supported not only by the graph shown in FIG. 9, but also by the meta-regression analysis results presented in [Table 2] and [Table 3]. [Table 2] showed that clinical background variables such as age, gender, and duration of depressive symptoms were closely related to changes in the slope of the reaction time curve, and [Table 3] showed that the interaction effect between the reaction time stability value and the degree of symptom improvement was statistically significant. Accordingly, the digital treatment response prediction device (100) can perform precise prediction that integrates multivariate factors rather than a simple average reaction time.
[0170] FIG. 10 is a diagram illustrating the process of a digital treatment response prediction device (100) according to one embodiment of the present invention classifying a responder group and a non-responder group by applying a Minimal Detectable Change (MDC) threshold and comparing and analyzing the EMA response time curves of each group.
[0171] The digital treatment response prediction device (100) can distinguish between a responding group and a non-responding group based on the Mean Difference Change (MDC) calculated from the response pattern model as shown in FIG. 9, and can also identify cases where improvement in depressive symptoms appears at a statistically significant level but no change is observed in the anxiety indicators by applying an anxiety symptom-related indicator in parallel under the same MDC standard as shown in FIG. 10. Through this, it is possible to differentiate between improvement in depression alone and a pattern of anxiety and depression together.
[0172] Figure 10(a) shows the results based on the Overall MDC (≤-3.81), and the responding group (n=14) exhibits a pattern of rapid adaptation and a sharp reduction in reaction time after the initial trial. Additionally, the responding group maintains a more consistent intra-individual reaction time (RT) pattern, with a small difference between the initial trial and optimal performance and low variability in reaction time across the entire curve. In contrast, the non-responding group (n=35) shows a large difference between the initial and later trials, a slow rate of improvement in reaction time, and relatively high variability in intra-individual reaction time. These differences are reflected not only in the curve graph on the left but also in the kernel density estimation (KDE) results on the right, confirming that the distribution of the responding group is more skewed to the left and concentrated.
[0173] Figure 10(b) shows the results based on the Adjusted MDC (≤-5.5). The responding group (n=4) exhibits extremely rapid initial adaptability and reaches a stabilized reaction time within a short period. During this process, the intra-individual reaction time (RT) is constant, and the difference between initial and optimal performance is minimized. In contrast, the non-responding group (n=45) has a flatter curve, limited improvement in reaction time, and lacks consistency in intra-individual reaction time. This is confirmed by the relatively wide distribution of the non-responding group in the KDE results on the right.
[0174] In this way, the digital treatment response prediction device (100) can capture the characteristics of the Responder as indicators such as rapid adaptability, a small difference between initial performance and optimal performance, and a more consistent intra-individual response time, as shown in FIG. 9. Therefore, the present invention goes beyond simply distinguishing between a responder group and a non-responder group by utilizing MDC criteria, and enables customized prediction and early treatment intervention based on the detailed response patterns of each patient.
[0176] FIG. 11 is a series of graphs illustrating the process of analyzing the gradual difference appearing in the EMA (Experience Sampling Method) reaction time curve according to the magnitude of the intervention effect according to one embodiment of the present invention.
[0177] Referring to Figure 11, the graph in the upper left shows the results for the elderly depression group. When the reaction time curves were classified according to the intervention effect size (delta value) based on the Geriatric Depression Scale (GDS), a distinct difference in initial reaction time was observed, and it was confirmed that the difference in the curves, particularly in the initial trial, acted as a distinguishing factor between the groups. However, since the Geriatric Depression Scale excludes items on somatic symptoms, it is interpreted that emotional and cognitive factors, rather than physical factors, explain the main differences.
[0178] The graph in the upper right corner shows the results for the general depression group. The General Depression Scale (CES-Dr) includes physical symptoms such as sleep, fatigue, appetite, and psychomotor items, and also incorporates suicide-related items. Consequently, the difference in the curves becomes more pronounced, suggesting that when physical symptoms and suicide items are included, differentiation between groups can be more clearly revealed in the EMA reaction time curve.
[0179] The graph in the lower left corner illustrates the results for the adult depression group. Since the Adult Depression Scale (PHQ-9) also includes physical items such as sleep, fatigue, appetite, and psychomotor symptoms, as well as items regarding suicide, the difference in the curve tends to be consistently maintained over a certain period. This demonstrates that in adult depression, the magnitude of the intervention effect is an important factor determining the slope of the EMA response curve and the difference in the time to stability response.
[0180] The graph in the bottom right illustrates the results for the anxiety symptom group. Differences in initial response times are particularly pronounced on the Anxiety Scale (BAI), suggesting that the response time during the initial administration is a key indicator for assessing treatment responsiveness in the anxiety symptom group. In other words, for anxiety patients, rapid adaptation during repeated EMA administrations can serve as a predictive variable.
[0181] The four graphs shown in Fig. 11 illustrate how the reaction time curve pattern changes depending on the composition of detailed items in the depression symptom assessment scale. For example, when suicide-related items were excluded from the geriatric depression scale, the difference in the initial slope of the curve was significantly mitigated; conversely, when only affect-related items were included, the convergence speed of the stability value (θc) was found to be distinctly improved. These results suggest that specific item groups have a differential impact on the reaction time patterns of the elderly and provide a basis for detailed modeling for future treatment targeting.
[0182] In summary, as illustrated in FIG. 11, the digital treatment response prediction device (100) analyzes the differences in the shape of the EMA response time curve for subgroups of depression and anxiety symptoms, and utilizes the difference in the initial trial as a key indicator for depression in the elderly, the difference based on whether somatic symptoms and suicide items are included for general and adult depression, and the difference in the adaptability of the initial response time for anxiety symptoms as key discriminant factors. Through this, the present invention can precisely identify the characteristics of each group according to the size of the intervention effect, which contributes to the establishment of customized treatment strategies in clinical settings.
[0184] The device described above may be implemented as a hardware component, a software component, and / or a combination of a hardware component and a software component. For example, the device and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. In addition, other processing configurations, such as parallel processors, are also possible.
[0185] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or command the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.
[0186] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.
[0187] Although the embodiments have been described above with reference to limited examples and drawings, those skilled in the art can make various modifications and variations from the description above. For example, suitable results can be achieved even if the described techniques are performed in a different order than described, and / or the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.
[0188] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.
[0189] Although specific embodiments have been described in the detailed description of this document, it will be obvious to those skilled in the art that various modifications are possible within the scope of this document.
Claims
Claim 1 A method for predicting a digital treatment response in an elderly patient with depression, performed by a digital treatment response prediction device, comprising: a step of collecting response time data for an ecological moment assessment performed by the elderly patient with depression through a user terminal; a step of generating a response pattern model of the elderly patient with depression by modeling the response pattern of the elderly patient with depression over time based on the response time data; and a step of predicting a digital treatment response in the elderly patient with depression using the response pattern model; wherein the step of predicting a digital treatment response in the elderly patient with depression comprises: a step of estimating the degree of improvement in depressive symptoms as good when the rate of change in response time derived from the response pattern model decreases below a preset reduction threshold value and the variance of response time converges below a preset stability threshold value. A method for predicting the response to digital treatment in elderly patients with depression using an ecological moment evaluation-based response time analysis, comprising: a step of differentially distinguishing between depressive symptoms and anxiety symptoms to determine whether depressive symptoms and anxiety symptoms coexist when the reduction threshold value and stability threshold value applied to the depressive symptom-related indicator regarding the reaction time change rate and the reaction time variance are applied to the anxiety symptom-related indicator, and when the anxiety symptom-related indicator does not satisfy the conditions for the reduction threshold value and the stability threshold value. Claim 2 A method for predicting the response to digital treatment in an elderly patient with depression using an ecological moment evaluation-based response time analysis, wherein the ecological moment evaluation includes items presented periodically or non-periodically through the user terminal during the daily life of the elderly patient with depression and response data thereof, and the response data includes self-report evaluation data that records the emotional, cognitive, and behavioral states at each response time in real time. Claim 3 A method for predicting digital treatment response in elderly patients with depression using ecological moment assessment-based response time analysis, comprising: a step of removing outliers by applying a quartile rule to the response time data prior to the step of generating a response pattern model of the elderly patient with depression in claim 1; and a step of performing standardization processing on the response time within an individual to control confounding effects arising from bias based on the configuration of response positions per item. Claim 4 A method for predicting the response to digital treatment in elderly patients with depression using ecological moment evaluation-based response time analysis, wherein, in claim 1, the response pattern model is generated by fitting the change in response time over time to an exponential decay curve. Claim 5 A method for predicting the response to digital treatment in elderly patients with depression using ecological moment evaluation-based response time analysis, wherein, in claim 4, the step of predicting the response to digital treatment in elderly patients with depression comprises: a step of extracting a plurality of parameters including the rate of change of response time, the rate of adaptation, or the variance of response time from the exponential decay curve. Claim 6 A method for predicting the response to digital treatment in an elderly patient with depression using an ecological moment evaluation-based response time analysis, wherein, in claim 5, the step of predicting the response to digital treatment in the elderly patient with depression comprises: a step of determining the possibility of symptom relief in the elderly patient with depression as favorable when the rate of change in response time decreases rapidly above a threshold value. Claim 7 A method for predicting a digital treatment response in an elderly patient with depression using an ecological moment evaluation-based response time analysis, wherein, in claim 5, the step of predicting a digital treatment response in an elderly patient with depression comprises: a step of classifying the patient as a treatment responder group when the adaptation speed is greater than or equal to a preset threshold value; and a step of classifying the patient as a non-responder group when the response time variance does not converge to or less than the threshold value. Claim 8 delete Claim 9 A digital treatment response prediction device comprises at least one processor; and a memory for storing instructions, wherein when the instructions are executed individually or collectively by the at least one processor, the processor performs: an operation of collecting response time data for an ecological moment assessment performed by an elderly patient with depression through a user terminal; an operation of generating a response pattern model of the elderly patient with depression by modeling the response pattern of the elderly patient with depression over time based on the response time data; and an operation of predicting the digital treatment response of the elderly patient with depression using the response pattern model; wherein the operation of predicting the digital treatment response of the elderly patient with depression includes an operation of estimating the degree of improvement of depressive symptoms as good when the rate of change in response time derived from the response pattern model decreases below a preset reduction threshold value and the variance of response time converges below a preset stability threshold value. A digital therapeutic response prediction device comprising: an operation of differentially distinguishing between depressive symptoms and anxiety symptoms to determine whether depressive symptoms and anxiety symptoms coexist when, for the above reaction time change rate and above reaction time variance, the decrease threshold value and the stability threshold value applied to the above depressive symptom-related indicator are applied to the anxiety symptom-related indicator, and the above anxiety symptom-related indicator does not satisfy the conditions for the decrease threshold value and the stability threshold value. Claim 10 A computer-readable recording medium that stores a computer program for performing a method for predicting the response to digital treatment in elderly patients with depression using an ecological moment assessment-based reaction time analysis according to any one of claims 1 to 7, combined with hardware.
Citation Information
Patent Citations
Method for predicting of depression and device for predicting of depression risk using the same
KR1020200001777A
Methods and systems for treating health conditions using digital prescription therapy
KR1020230164001A
System for screening depression and monitoring symptom using digital contents and method thereof
KR1020250072334A
