SYSTEMS AND METHODS FOR UNOBSTRUCTED DIGITAL HEALTH ASSESSMENT

Description

Priority Claim

[0001] The present application claims priority from: Indian Patent Application No. 201821028541, filed on 30th July, 2018.TECHNICAL FIELD

[0002] The disclosure herein generally relates to health assessment, and, more particularly, to a platform that facilitates screening of high risk subjects by monitoring their daily lives in an unobtrusive manner.BACKGROUND

[0003] Existing healthcare systems are mostly based on reactive medicine whereby person goes to a medical practitioner only when she is symptomatic. Another approach relies on periodic health checkups in an attempt to discover medical conditions at an early stage. Both the approaches have some issues. The first approach suffers from the fact that by the time some diseases become symptomatic, considerable damage may have already been done. An issue with the second approach is that since patient is asymptotic, the checkups may result in no diagnosis at all; the fallout rate is also high owing to people's busy schedules. Patent publication US 2014 / 0046683 discloses methods and systems useful for characterizing clinical outcomes of a subject. Provided herein includes computer-assessed methods, medical information systems, and computer-readable instructions that can aid an end-user in diagnosis, prognosis, and treatment of a clinical outcome (Abstract).SUMMARY

[0004] Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems.

[0005] In an aspect, there is provided a processor implemented methodis as defined in claim 1.

[0006] In another aspect, there is provided a system as defined in claim 6 .

[0007] In yet another aspect, there is provided a computer program product comprising a non-transitory computer readable medium having a computer readable program embodied therein, wherein the computer readable program, when executed on a computing device as defined in claim 11.

[0008] In an embodiment of the present disclosure, the one or more hardware processors are further configured to obtain values associated with a plurality of metadata features by (i) directly deriving the values using corresponding measurement devices or (ii) estimating the values based on the domain knowledge of the high risk subjects under consideration.

[0009] In an embodiment of the present disclosure, the one or more hardware processors are further configured to randomly sample the two clusters based on a mean and a standard deviation associated with the obtained skewed normal distribution of the high risk subjects.

[0010] In an embodiment of the present disclosure, the neural network uses a multi-layer perceptron with at least two hidden layers and a fully connected input and output layer.

[0011] In an embodiment of the present disclosure, the one or more hardware processors are further configured to identify a level of deterioration of health of each of the high risk subjects in the treatment class by: monitoring physical activity levels and physiological measurements of the high risk subjects from the treatment class; classifying each of the high risk subjects into one of a plurality of pre-determined classes illustrative of health assessment thereof using a computational model and a correlation between the monitored physical activity levels and the physiological measurements; predicting a normalized value for each of the physiological measurements of interest using a Hidden Markov Model (HMM); and computing a measure of deviation from the predicted normalized value using an actual normalized value obtained from the monitoring physiological measurements to assess deviation from a healthy condition for each of the high risk subjects and identifying an associated metadata feature as a bio-marker for further assessment of a corresponding high risk subject.

[0012] In an embodiment of the present disclosure, the one or more hardware processors are further configured to eliminate local outliers in the physiological measurements using a Local Outlier Filter (LOF) algorithm to obtain filtered physiological measurements; perform a trend analyses, of the monitored physical activity levels and the filtered physiological measurements, using AutoRegressive Integrated Moving Average (ARIMA); and trigger an alarm when a trend is negative with a slope greater than a pre-defined threshold.

[0013] In an embodiment of the present disclosure, the one or more hardware processors are further configured to compute a measure of deviation from the predicted value by analyzing feedback pertaining to intensity associated with the monitored physical activity levels from the high risk subjects.

[0014] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles. FIG.1 illustrates an exemplary block diagram of a system for unobtrusive digital health assessment, in accordance with an embodiment of the present disclosure. FIG.2A through FIG.2D illustrate exemplary flow charts for a computer implemented method for unobtrusive digital health assessment, in accordance with an embodiment of the present disclosure. FIG.3A and FIG.3B illustrate an exemplary control class and an exemplary treatment class respectively of high risk subjects obtained in the art. FIG.4A and FIG.4B illustrate a skewed normal distribution of the high risk subjects in a control class and a treatment class respectively of high risk subjects obtained in accordance with an embodiment of the present disclosure. FIG.5 illustrates a schematic representation of a high level implementation of the method illustrated in FIG.2A through FIG.2D for unobtrusive digital health assessment in accordance with an embodiment of the present disclosure. FIG.6A and FIG.6B illustrate heart rate values and breathing signal power respectively experienced by an exemplary subject undergoing a treadmill experiment at a speed of 4.2 kmph. FIG.7A and FIG.7B illustrate heart rate values and breathing signal power respectively experienced by the exemplary subject of FIG.6A and FIG.6B undergoing a treadmill experiment at a speed of 5.4 kmph. FIG.8A and FIG.8B illustrate heart rate values and breathing signal power respectively experienced by the exemplary subject of FIG.6A and FIG.6B undergoing a treadmill experiment at a speed of 7.8 kmph. FIG.9A and FIG.9B illustrate heart rate values and breathing signal power respectively experienced by the exemplary subject of FIG.6A and FIG.6B undergoing a treadmill experiment at a speed of 9.0 kmph. DETAILED DESCRIPTION OF EMBODIMENTS

[0016] Exemplary embodiments are described with reference to the accompanying drawings. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. Wherever convenient, the same reference numbers are used throughout the drawings to refer to the same or like parts. While examples and features of disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the scope of the disclosed embodiments. It is intended that the following detailed description be considered as exemplary only, with the true scope being indicated by the following claims.

[0017] Systems and methods of the present disclosure enable screening of diseases in high risk subjects by monitoring their daily life. In the context of the present disclosure, the expression 'high risk subjects' are subjects with predisposing systemic conditions including neuropathy, peripheral arterial disease, diabetes mellitus, proneness to infection, autoimmune disease and immunocompromised rendering them more likely than others to get a particular disease. The present disclosure enables a platform that integrates physical characteristics, lifestyle habits and prevailing medical conditions with monitored physical activities and physiological measurements to assess health of high risk subjects. In the context of the present disclosure, the physical characteristics includes height, weight, gender, ethnicity, age and the like. Likewise, lifestyle habits may include smoking, drinking, exercising regularly, over-eating, and the like. Further, in the context of the present disclosure, prevailing medical conditions includes diabetes, hypertension, cardiac illness, anemia, and the like.

[0018] Referring now to the drawings, and more particularly to FIGS. 1 through 9B, where similar reference characters denote corresponding features consistently throughout the figures, there are shown preferred embodiments and these embodiments are described in the context of the following exemplary system and method.

[0019] FIG.1 illustrates an exemplary block diagram of a system 100 for for unobtrusive digital health assessment in accordance with an embodiment of the present disclosure. The system 100 includes one or more processors 104, communication interface device(s) or input / output (I / O) interface(s) 106, and one or more data storage devices or memory 102 operatively coupled to the one or more processors 104. The one or more processors 104 that are hardware processors can be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, graphics controllers, logic circuitries, and / or any devices that manipulate signals based on operational instructions. Among other capabilities, the processor(s) are configured to fetch and execute computer-readable instructions stored in the memory. In the context of the present disclosure, the expressions 'processors' and 'hardware processors' are used interchangeably. The system 100 can be implemented in a variety of computing systems, such as laptop computers, notebooks, hand-held devices, workstations, mainframe computers, servers, a network cloud and the like.

[0020] The I / O interface(s) 106 can include a variety of software and hardware interfaces, for example, a web interface, a graphical user interface, and the like and can facilitate multiple communications within a wide variety of networks N / W and protocol types, including wired networks, for example, LAN, cable, etc., and wireless networks, such as WLAN, cellular, or satellite. The I / O interface(s) can include one or more ports for connecting a number of devices to one another or to another server.

[0021] The memory 102 includes any computer-readable medium known in the art including, for example, volatile memory, such as static random access memory (SRAM) and dynamic random access memory (DRAM), and / or nonvolatile memory, such as read only memory (ROM), erasable programmable ROM, flash memories, hard disks, optical disks, and magnetic tapes. The one or more modules (not shown) of the system 100 can be stored in the memory 102.

[0022] FIG.2 is an exemplary flow diagram illustrating a computer implemented method for unobtrusive digital health assessment. The system 100 includes one or more data storage devices or memory 102 operatively coupled to the one or more processors 104 and is configured to store instructions configured for execution of steps of the method 200 by the one or more processors 104. The steps of the method 200 will now be explained in detail with reference to the components of the system 100 of FIG. 1. Although process steps, method steps, techniques or the like may be described in a sequential order, such processes, methods and techniques may be configured to work in alternate orders. In other words, any sequence or order of steps that may be described does not necessarily indicate a requirement that the steps be performed in that order. The steps of processes described herein may be performed in any order practical. Further, some steps may be performed simultaneously.

[0023] Accordingly, in an embodiment of the present disclosure, the one or more processors 104 are configured to obtain, at step 202, values associated with a plurality of metadata features, wherein the plurality of metadata features constitute domain knowledge of high risk subjects under consideration and comprise one or more of physical characteristics, lifestyle habits and prevailing medical conditions. The step of obtaining values comprise directly deriving the values using corresponding measurement devices. Alternatively, the values may be estimated based on the domain knowledge of the high risk subjects under consideration.

[0024] In an embodiment of the present disclosure, the one or more processors 104 are configured to generate, at step 204, groups of the high risk subjects under consideration based on a plurality of combinations of the obtained plurality of metadata features. For instance, there may be a group of age 40+ high risk subjects, a group of age 40+ high risk subjects with weight 80+kg, a group of age 40+ high risk subjects with height more than 6 feet, a group of age 40+ high risk subjects with weigh 80+kg and having cardiac issues, and the like.

[0025] Say, there is a dataset of subjects, some of which have a certain disease and some who do not have the disease. Let these two classes be denoted as 1 and 0, respectively. Metadata features of the high risk subjects in both classes are obtained. For instance, say age information is obtained. Given the disease population (class 1), it is critical to decide whether the control class (class 0) is close to the treatment class, or far, based on the distribution of the metadata features. This involves generating a control class so that the correlation coefficient between the distributions of the metadata features of both classes is close to 1. FIG.3Aand FIG.3B illustrate an exemplary control class and an exemplary treatment class respectively of high risk subjects obtained in the art, wherein age information as a metadata feature was simulated for heart disease patients from different countries. Originally, the treatment and control classes were of the size 480 and 1528 and the distributions are as illustrated in FIG. 3A and FIG.3B respectively. It is to be noted that the control class distribution resembles a decaying power law relation whereas the treatment class is the opposite giving an impression that age is a driving factor for the disease. Grouping of subjects automatically based on metadata features is challenging because unsupervised grouping is performed using clustering techniques and metadata features are often categorical or subjective.

[0026] In accordance with the present disclosure, the system 100 and method 200 explained hereinafter facilitate automatic generation of a treatment class and a control class which can ultimately result in identifying bio-markers for further assessment of the high risk subjects. Accordingly, in an embodiment, the one or more processors 104 are configured to iteratively obtain, at step 206, a skewed normal distribution of the high risk subjects to generate the treatment class and the control class for each of the groups generated at step 204. The step 206 addresses the issue seen in FIG.3A and FIG.3B explained herein above. Histogram information of binned data of both classes pertaining to the distribution of the metadata features is used. The control class may be generated such that a ratio of the bins in its histogram is close to the ratio of bins in the treatment population histogram. It is mathematically shown that for large treatment and control classes (size <=50), the correlation between the classes is close to 1.

[0027] Particularly, as part of step 206, in an embodiment, firstly a fuzzy membership function based on a neural network is generated at step 206a and two feature classes are derived for each of the obtained metadata features. For instance, the classes is of young / old for age as a metadata feature, the classes is of tall / short for height as a metadata feature, and the like. In an embodiment, the neural network uses a multi-layer perceptron with at least two hidden layers and a fully connected input and output layer. At step 206b, a plurality of normalized values between 0 and 1 are derived for each of the high risk subjects in the two derived feature classes, wherein a normalized value in the plurality of normalized values corresponds to a metadata feature. For instance, hypothetically speaking, if thin is 0 and fat is 1 for body type as a metadata feature and say if old is 0 and young is 1, a high risk subject who is thin and old have normalized values like 0.2 thin, 0.8 old. At step 206c, a Manhattan distance between every pair of subjects amongst the high risk subjects is obtained using the plurality of normalized values derived at step 206b. Two clusters of the high risk subjects are then generated at step 206d based on the obtained Manhattan distance and using a fuzzy C-means clustering method such that each of the two clusters have equivalent normalized values. The fuzzy C-means clustering ensures inter-cluster distance is high whereas intra-cluster distance is low. The two clusters generated at step 206d are then randomly sampled at step 206e to create the control class and the treatment class such that both the classes comprise an equivalent number of high risk subjects from each of the two clusters. FIG.4A and FIG.4B illustrate a skewed normal distribution of the high risk subjects in a control class and a treatment class respectively of high risk subjects obtained in accordance with an embodiment of the present disclosure. In an embodiment, the random sampling is based on a mean and a standard deviation associated with the obtained skewed normal distribution of the high risk subjects. Once the treatment and control classes are created, the plurality of normalized values associated with each of the high risk subjects are reconstructed at step 206f to obtain actual values corresponding to the associated metadata features for each of the high risk subjects in the control class and the treatment class. In an embodiment, the reconstruction involves correlating an identifier of a subject with a meta-dictionary.

[0028] If the normal for a particular subject is personalized and if an attempt is made to derive abnormal with respect to the normal, it is to be noted that outliers and abnormal readings due to temporary health problems may be captured. This anomaly is to be averted by taking a longer measurement and running a Local Outlier Filter (LOF) algorithm over a time series. However, quantifying deviation to gauge a level of deterioration of health is a challenge. In accordance with an embodiment of the present disclosure, the method 200 further comprises a step 208, wherein the one or more processors 104 are configured to identify the level of deterioration of health of each of the high risk subjects in the treatment class. Towards this, firstly physical activity levels and physiological measurements of the high risk subjects from the treatment class is monitored at step 208a. For instance, how long and what distance has the high risk subject walked today is measured. Such monitoring may be performed using wearable devices and allied mobile or stationary gateways.

[0029] In accordance with an embodiment of the present disclosure, each of the high risk subjects is then be classified at step 208b into one of a plurality of pre-determined classes illustrative of health assessment of the high risk subject. This classification is performed using a computational model and a correlation between the monitored physical activity levels and the physiological measurements. In an embodiment, the computation model is based on New York Heart Association (NYHA) guidelines wherein 4 classes are identified to depict stages of heart failure. As per NYHA guidelines, Class I is mapped to a condition that ordinary physical activity does not cause undue fatigue, palpitations, dyspnea and / or angina; Class II is mapped to a condition that ordinary physical activity does cause undue fatigue, palpitations, dyspnea and / or angina; Class III is mapped to a condition that less than ordinary physical activity causes undue fatigue, palpitations, dyspnea and / or angina; and Class IV is mapped to a condition that fatigue, palpitations, dyspnea and / or angina occurs at rest. In an aspect of the present disclosure, a metadata feature fatigue is modelled in the form of I = f(MET, W, T, LPA, G, A, N), wherein I represents intensity as a function of intensity associated with the monitored physical activity levels (MET), W represents body weight of the high risk subject under consideration, T represents Spell's duration, LPA represents level of physical activity in daily life of the high risk subject under consideration, G represents gender of the high risk subject under consideration, A represents age spectrum and N represents a normalization constant.

[0030] In accordance with an embodiment of the present disclosure, at step 208c, a normalized value for each of the physiological measurements is predicted using a Hidden Markov Model (HMM). The prediction using HMM is performed using stage wise prediction, the parameters being MET, breathing power change, heart rate change, breathing rate change and time taken to return to a Basal heart rate (normal heart rate at rest for the high risk subject under consideration). Accordingly, Class II is identified if a high risk subject gets tired after say 6 minutes (to be normalized) of walking.

[0031] In accordance with an embodiment, at step 208d, a measure of deviation from the predicted normalized value using an actual normalized value obtained from the monitoring physiological measurements is computed. The deviation is computed as a mean deviation wherein lower quadrant values is considered and upper quadrant values is only be considered if no lower quadrant values are available. The measure of deviation is then be used to assess deviation from a healthy condition for each of the high risk subjects and to identify an associated metadata feature as a bio-marker for further assessment of a corresponding high risk subject. For instance, in Class II ordinary physical activity does cause undue fatigue, palpitations, dyspnea and / or angina. If the subject does get tired after an ordinary physical activity like say after 2 minutes of walking, the high risk subject is classified into Class II. Further, a feedback is requested, from the high risk subject pertaining to MET is analyzed and whether he does feel tired after 2 minutes of walking as measured. Also, a metadata feature associated with the tiredness is identified as a bio-marker for further assessment of a corresponding high risk subject.

[0032] In an embodiment, local outliers is eliminated, at step 208e, in the physiological measurements using a Local Outlier Filter (LOF) algorithm to obtain filtered physiological measurements. A trend analyses is performed, at step 208f, of the monitored physical activity levels and the filtered physiological measurements using AutoRegressive Integrated Moving Average (ARIMA). For instance, some days the high risk subject have a headache leading to spikes in the trend analyses which clearly need to be ignored. Each of the high risk subjects have different energy levels at different times of the day or have seasonal variations leading to a trend. Over time, a subject gets more tired and such long term trend is identified using ARIMA. At step 208g, an alarm is triggered when the trend is negative with a slope greater than a pre-defined threshold.

[0033] Thus, in accordance with the present disclosure, systems and methods of the present disclosure finds application in both pre-op and post-op scenarios. FIG.5 illustrates a schematic representation of a high level implementation of the method illustrated in FIG.2A through FIG.2D for unobtrusive digital health assessment in accordance with an aspect of the present disclosure. The physiological parameters when monitored before and after an activity for a particular MET is used to assess the levels of cardiopulmonary fatigue in a high risk subject under consideration who otherwise appears asymptomatic. The fatigue levels of the high risk subject under consideration is then be normalized considering other high risk subjects in the same group as explained above. If the normalized level of fatigue is higher than a major percentage of population in the group, the high risk subject is marked for further assessment based on fatigue level being identified as a bio-marker. Such marked high risk subjects is monitored longitudinally over time to check if the fatigue levels are trending negatively to trigger a timely alarm. If the fatigue levels are not trending negatively, the pre-op monitoring continues assessing the health of the high risk subject in an unobtrusive manner as explained. Conventionally, identifying a bio-marker involve insights derived through obtrusive medical tests and correlating such insights with activities of daily living (ADL) is a challenge. Post-op care of a high risk subject is more vital and risk prone. Accordingly, the systems and methods of the present disclosure comprises accurate monitoring of movement patterns using devices such as Kinect and the physiological parameters is also checked using sophisticated medical devices (for instance, measuring of blood oxygen saturation levels SpO2). In an aspect of the present disclosure, a persuasion engine (not shown) is comprised in the system 100 to involve a care taker for the high risk subject under consideration for ensuring post-op care and requirements are adhered without fail.

[0034] Experimental validation for the method of the present disclosure was performed with high risk subjects walking on a treadmill at different speeds and elevations. Physiological measurements were obtained along with recovery time before and after the walking session. FIG.6A and FIG.6B illustrate heart rate values and breathing signal power respectively experienced by an exemplary subject undergoing a treadmill experiment at a speed of 4.2 kmph and FIG.7A and FIG.7B illustrate heart rate values and breathing signal power respectively experienced by the exemplary subject of FIG.6A and FIG.6B undergoing a treadmill experiment at a speed of 5.4 kmph. It was noted that there was no substantial change in heart rate, no noticeable fatigue was experienced by the high risk subject and breathing power shot up momentarily but settled down very quickly (very less recovery time.

[0035] FIG.8A and FIG.8B illustrate heart rate values and breathing signal power respectively experienced by the exemplary subject of FIG.6A and FIG.6B undergoing a treadmill experiment at a speed of 7.8 kmph and FIG.9A and FIG.9B illustrate heart rate values and breathing signal power respectively experienced by the exemplary subject of FIG.6A and FIG.6B undergoing a treadmill experiment at a speed of 9.0 kmph. It was noted that there was substantial change in heart rate, shortness of breath and fatigue was reported by the high risk subject under consideration and breathing power increased after the session and settled down gradually reflecting more recovery time. This observation clearly indicates that although the high risk subject under consideration is asymptomatic, further assessment is necessary to diagnose a medical condition, if any, that need attention.

[0036] The written description describes the subject matter herein to enable any person skilled in the art to make and use the embodiments. The scope of the subject matter embodiments is defined by the claims.

[0037] It is to be understood that the scope of the protection is extended to such a program and in addition to a computer-readable means having a message therein; such computer-readable storage means contain program-code means for implementation of one or more steps of the method, when the program runs on a server or mobile device or any suitable programmable device. The hardware device can be any kind of device which can be programmed including e.g. any kind of computer like a server or a personal computer, or the like, or any combination thereof. The device may also include means which could be e.g. hardware means like e.g. an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, e.g. an ASIC and an FPGA, or at least one microprocessor and at least one memory with software modules located therein. Thus, the means can include both hardware means and software means. The method embodiments described herein could be implemented in hardware and software. The device may also include software means. Alternatively, the embodiments may be implemented on different hardware devices, e.g. using a plurality of CPUs.

[0038] The embodiments herein can comprise hardware and software elements. The embodiments that are implemented in software include but are not limited to, firmware, resident software, microcode, etc. The functions performed by various modules described herein may be implemented in other modules or combinations of other modules. For the purposes of this description, a computer-usable or computer readable medium can be any apparatus that can comprise, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

[0039] The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Also, the words "comprising," "having," "containing," and "including," and other similar forms are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items, or meant to be limited to only the listed item or items. It must also be noted that as used herein and in the appended claims, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise.

[0040] Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term "computer-readable medium" should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, nonvolatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.

Claims

1. A processor implemented method (200) comprising: obtaining values associated with a plurality of metadata features, wherein the plurality of metadata features constitute domain knowledge of high risk subjects under consideration and comprise physical characteristics, lifestyle habits and prevailing medical conditions (202), wherein the high risk subjects are subjects with predisposing systemic conditions generating groups of the high risk subjects under consideration based on a plurality of combinations of the obtained plurality of metadata features (204); and iteratively obtaining a skewed normal distribution of the high risk subjects to generate a treatment class and a control class for each of the generated groups (206), the step of obtaining a skewed normal distribution comprising: generating a fuzzy membership function based on a neural network and deriving two feature classes for each of the obtained metadata features (206a); deriving a plurality of normalized values between 0 and 1 for each of the high risk subjects in the two derived feature classes, wherein a normalized value in the plurality of normalized values corresponds to a metadata feature (206b); obtaining a Manhattan distance between every pair of subjects amongst the high risk subjects using the plurality of normalized values (206c); generating two clusters of the high risk subjects based on the obtained Manhattan distance and using a fuzzy C-means clustering method such that each of the two clusters have equivalent normalized values (206d); randomly sampling the two clusters to create the control class and the treatment class such that the control class and the treatment class comprises an equivalent number of high risk subjects from each of the two clusters (206e); reconstructing the plurality of normalized values associated with each of the high risk subjects by correlating an identifier of the high risk subject with a meta-dictionary to obtain actual values corresponding to the associated metadata features for each of the high risk subjects in the control class and the treatment class (206f); and identifying a level of deterioration of health of each of the high risk subjects in the treatment class (208) by: monitoring physical activity levels related to a time spent and a distance achieved while performing the physical activity measured using a wearable device and physiological measurements of the high risk subjects from the treatment class (208a); classifying each of the high risk subjects into one of a plurality of pre-determined classes depicting stages of a disease using a computational model and a correlation between the monitored physical activity levels and the physiological measurements (208b); predicting a normalized value for each of the physiological measurements of interest using a Hidden Markov Model (HMM) (208c); computing a measure of deviation from the predicted normalized value using an actual normalized value obtained from the monitoring physiological measurements to assess deviation from a healthy condition for each of the high risk subjects and identifying an associated metadata feature as a bio-marker for further assessment of a corresponding high risk subject (208d); eliminating local outliers in the physiological measurements using a Local Outlier Filter (LOF) algorithm to obtain filtered physiological measurements (208e); performing a trend analyses, of the monitored physical activity levels and the filtered physiological measurements and identifying a long term trend using AutoRegressive Integrated Moving Average (ARIMA) (208f); and triggering an alarm when the trend is negative with a slope greater than a pre-defined threshold (208g).

2. The processor implemented method of claim 1, wherein the step of obtaining values associated with a plurality of metadata features comprises (i) directly deriving the values using corresponding measurement devices or (ii) estimating the values based on the domain knowledge of the high risk subjects under consideration.

3. The processor implemented method of claim 1, wherein the neural network uses a multi-layer perceptron with at least two hidden layers and a fully connected input and output layer.

4. The processor implemented method of claim 1, wherein the random sampling is based on a mean and a standard deviation associated with the obtained skewed normal distribution of the high risk subjects.

5. The processor implemented method of claim 1, wherein computing a measure of deviation from the predicted value is preceded by analyzing feedback pertaining to intensity associated with the monitored physical activity levels from the high risk subjects.

6. A system (100) comprising: one or more data storage devices (102) operatively coupled to one or more hardware processors (104) and configured to store instructions configured for execution by the one or more hardware processors (104) to: obtain values associated with a plurality of metadata features, wherein the plurality of metadata features constitute domain knowledge of high risk subjects under consideration and comprise physical characteristics, lifestyle habits and prevailing medical conditions, wherein high risk subjects are subjects with predisposing systemic conditions; generate groups of the high risk subjects under consideration based on a plurality of combinations of the obtained plurality of metadata features; and iteratively obtain a skewed normal distribution of the high risk subjects to generate a treatment class and a control class for each of the generated groups, wherein the skewed normal distribution is obtained by: generating a fuzzy membership function based on a neural network and deriving two feature classes for each of the obtained metadata features; deriving a plurality of normalized values between 0 and 1 for each of the high risk subjects in the two derived feature classes, wherein a normalized value in the plurality of normalized values corresponds to a metadata feature; obtaining a Manhattan distance between every pair of subjects amongst the high risk subjects using the plurality of normalized values; generating two clusters of the high risk subjects based on the obtained Manhattan distance and using a fuzzy C-means clustering method such that each of the two clusters have equivalent normalized values; randomly sampling the two clusters to create the control class and the treatment class such that the control class and the treatment class comprises an equivalent number of high risk subjects from each of the two clusters; reconstructing the plurality of normalized values associated with each of the high risk subjects by correlating an identifier of the high risk subject with a meta-dictionary to obtain actual values corresponding to the associated metadata features for each of the high risk subjects in the control class and the treatment class; and identifying a level of deterioration of health of each of the high risk subjects in the treatment class (208) by: monitoring physical activity levels related to a time spent and a distance achieved while performing the physical activity measured using a wearable device and physiological measurements of the high risk subjects from the treatment class (208a); classifying each of the high risk subjects into one of a plurality of pre-determined classes depicting stages of a disease using a computational model and a correlation between the monitored physical activity levels and the physiological measurements (208b); predict a normalized value for each of the physiological measurements of interest using a Hidden Markov Model (HMM); compute a measure of deviation from the predicted normalized value using an actual normalized value obtained from the monitoring physiological measurements to assess deviation from a healthy condition for each of the high risk subjects and identifying an associated metadata feature as a bio-marker for further assessment of a corresponding high risk subject; eliminate local outliers in the physiological measurements using a Local Outlier Filter (LOF) algorithm to obtain filtered physiological measurements; perform a trend analyses, of the monitored physical activity levels and the filtered physiological measurements and identify a long term trend using AutoRegressive Integrated Moving Average (ARIMA); and trigger an alarm when the trend is negative with a slope greater than a pre-defined threshold.

7. The system (100) of claim 8, wherein the one or more hardware processors (104) are further configured to obtain values associated with a plurality of metadata features by (i) directly deriving the values using corresponding measurement devices or (ii) estimating the values based on the domain knowledge of the high risk subjects under consideration.

8. The system (100) of claim 8, wherein the neural network uses a multi-layer perceptron with at least two hidden layers and a fully connected input and output layer.

9. The system (100) of claim 8, wherein the one or more hardware processors (104) are further configured to randomly sample the two clusters based on a mean and a standard deviation associated with the obtained skewed normal distribution of the high risk subjects.

10. The system (100) of claim 6, wherein the one or more hardware processors (104) are further configured to compute a measure of deviation from the predicted value by analyzing feedback pertaining to intensity associated with the monitored physical activity levels from the high risk subjects.

11. A computer program product comprising a non-transitory computer readable medium having a computer readable program embodied therein, wherein the computer readable program, when executed on a computing device, causes the computing device to: obtain values associated with a plurality of metadata features, wherein the plurality of metadata features constitute domain knowledge of high risk subjects under consideration and comprise physical characteristics, lifestyle habits and prevailing medical conditions, wherein high risk subjects are subjects with predisposing systemic conditions; generate groups of the high risk subjects under consideration based on a plurality of combinations of the obtained plurality of metadata features; iteratively obtain a skewed normal distribution of the high risk subjects to generate a treatment class and a control class for each of the generated groups, wherein the skewed normal distribution is obtained by: generating a fuzzy membership function based on a neural network and deriving two feature classes for each of the obtained metadata features; deriving a plurality of normalized values between 0 and 1 for each of the high risk subjects in the two derived feature classes, wherein a normalized value in the plurality of normalized values corresponds to a metadata feature; obtaining a Manhattan distance between every pair of subjects amongst the high risk subjects using the plurality of normalized values; generating two clusters of the high risk subjects based on the obtained Manhattan distance and using a fuzzy C-means clustering method such that each of the two clusters have equivalent normalized values; randomly sampling the two clusters to create the control class and the treatment class such that the control class and the treatment class comprises an equivalent number of high risk subjects from each of the two clusters; reconstructing the plurality of normalized values associated with each of the high risk subjects by correlating an identifier of the high risk subject with a meta-dictionary to obtain actual values corresponding to the associated metadata features for each of the high risk subjects in the control class and the treatment class; and identify a level of deterioration of health of each of the high risk subjects in the treatment class by performing: monitoring physical activity levels related to a time spent and a distance achieved while performing the physical activity measured using a wearable device and physiological measurements of the high risk subjects from the treatment class; classifying each of the high risk subjects into one of a plurality of pre-determined classes depicting stages of a disease using a computational model and a correlation between the monitored physical activity levels and the physiological measurements; predicting a normalized value for each of the physiological measurements of interest using a Hidden Markov Model (HMM); computing a measure of deviation from the predicted normalized value using an actual normalized value obtained from the monitoring physiological measurements to assess deviation from a healthy condition for each of the high risk subjects and identifying an associated metadata feature as a bio-marker for further assessment of a corresponding high risk subject; eliminating local outliers in the physiological measurements using a Local Outlier Filter (LOF) algorithm to obtain filtered physiological measurements; performing a trend analyses, of the monitored physical activity levels and the filtered physiological measurements and identifying a long term trend using AutoRegressive Integrated Moving Average (ARIMA); and triggering an alarm when the trend is negative with a slope greater than a pre-defined threshold.