An adult early warning system to predict the risk of clinical deterioration of patients and a method thereof

Abstract

The present disclosure proposes an adult early warning system (EWS) (100) for predicting clinical deterioration and mortality risks in patients. The EWS (100) integrates via API, utilizing REST API and HTTPS for secure communication. It features an input module (112) for authenticating and validating data, which is transmitted through a network (102) to a server (104). The server (104) processes data through data collection, standardization, and pre-processing modules (122, 124) before it is analyzed by the prediction module (116). Results are aligned with the clinical pathway module (118) and presented by the risk prediction response and patient overview dashboard module (120), which provides risk scores, and graphical representations of vitals, lab results, and ECG data. The EWS (100) supports clinicians by enhancing decision-making and adhering to ISO standards for patient safety and reliability.

Description

4. DESCRIPTION:Field of the invention:

[0001] The present disclosure relates to the technical field of health monitoring systems, with a particular emphasis on an adult early warning system designed to identify patients at risk of clinical deterioration and mortality. This system predicts risk scores at intervals of 1 hour, 4 hours, 8 hours, 12 hours, and 24 hours using a sophisticated triphasic machine learning model.Background of the invention:

[0002] Doctors and physicians continuously monitor patient conditions in acute care settings, interpreting signs, symptoms, and lab and Imaging test results to make timely diagnoses and treatment decisions. Rapid response is critical in these settings, as patient conditions can deteriorate quickly without immediate intervention. The Adult Early Warning System (EWS) can be instrumental in these situations due to its ability to predict clinical deterioration early.

[0003] Take for example, consider a patient with sepsis in the ICU. While initially stabilized, the patient remains at high risk for sudden deterioration. If the patient's temperature spikes and their white blood cell count rises, indicating a potential worsening of the infection, the EWS can detect these changes before the patient exhibits overt clinical signs. The system provides a graphical trend of these parameters and identifies contributing factors, prompting healthcare providers to intervene swiftly. This might involve administering antibiotics, increasing fluid resuscitation, or preparing for more aggressive supportive measures.

[0004] In another scenario, consider a patient in the ICU recovering from a major surgery. The recovery process is often complex and the patient is at risk of post-operative complications, some of which can be life-threatening if not detected early. Here, the EWS can play a crucial role in monitoring the patient's condition. For example, if the patient's respiration rate starts increasing and oxygen saturation begins to decrease, these could be early signs of a potential lung complication. The EWS, by detecting these changes early,2allows for immediate intervention such as supplemental oxygen or respiratory support, thereby preventing further deterioration.

[0005] Similarly, in an inpatient setting such as a high-dependency unit, the EWS enables physicians to anticipate clinical deterioration in patients recovering from acute illnesses or procedures. If the EWS indicates a heightened risk of deterioration, healthcare providers can promptly Implement preventive measures. This could include closer monitoring, additional diagnostic tests, or preemptive therapeutic interventions. By leveraging the predictive capabilities of the EWS, clinicians can tailor interventions to each patient's risk profile, aiming to prevent or minimize clinical deterioration and improve overall patient outcomes.

[0006] Traditional health monitoring systems are primarily designed to track real-time vital signs, including heart rate, blood pressure, temperature, and oxygen saturation. These systems alert healthcare providers when these parameters deviate from the normal ranges, providing critical information about a patient's immediate health status. However, these systems are largely reactive, offering information only when a patient's condition has already begun to deteriorate. They lack the predictive capabilities required to foresee potential health risks and enable preemptive interventions.

[0007] Conventional health monitoring systems typically employ static algorithms based on standardized physiological thresholds, generally developed using Western data. These algorithms do not consider individual patient history or demographic factors, often resulting in a lack of cultural and demographic specificity for diverse patient populations. This limited approach does not account for the complexity and variability of human health conditions, thus limiting the accuracy and applicability of these systems.

[0008] Most traditional health monitoring systems are rigid, with fixed parameters and thresholds. They lack the ability to adapt to new data or different patient conditions, and do not provide a comprehensive view of a patient's health status. This lack of adaptability and holistic view can lead to delayed or inaccurate diagnosis and interventions, potentially impacting patient outcomes.

[0009] In contrast to conventional systems, EWS extends beyond real-time monitoring of vital signs. The Early Warning System (EWS) is an Innovative Al-enabled healthcare system that offers a significantly advanced approach to traditional health monitoring. Utilizing machine learning and data from Indian medical records, EWS operates with an accuracy range between 83% and 92%. The system employs a unique three-phase model that integrates a vast quantity of ethically obtained, deanonymized data from various Apollo hospital regions such as Chennai, Bangalore, Hyderabad, Delhi, Mumbai, Kolkata, Nashik, Bhubaneswar, over different time periods. The retrospective Phase 1 covers January to October 2023, while Phase 2 and 3 span from January to December 2022.

[0010] During the first phase, XGB model analyzed vital signs and demographic data from over 700 million data points. After excluding missing data, the final sample size was reduced to 32,000 patients. This model provides a risk score and categorizes risk levels with an accuracy of 83%. The second phase introduces a neural network model that evaluates data from over 145,000 patients, broadening its input to include more comprehensive vital signs, comorbidities, and symptoms, and is designed to predict both mortality and clinical deterioration with an accuracy rate of 84%.

[0011] In Phase 3 integrates an XGB model using data from over 145,000 patients. It incorporates vital signs, comorbidities, and a comprehensive set of laboratory parameters for a more holistic prediction of mortality or clinical deterioration. This phase shows ab accuracy rate of 92%. Throughout all phases, the EWS provides numerical confidence levels and risk categorization, serving as an essential system for caregivers to make informed decisions. When the current data is not available, the API and Algorithm shall use imputation techniques like forward fill (last observed value).

[0012] The EWS uses event rates of 8.6% for clinical deterioration and 2.8% for mortality, derived from test and validation data. Propensity match scores have been obtained to address potential selection bias, unmeasured confounders, and heterogeneity of comorbidities. As a versatile system, EWS can be integrated into both inpatient and remote care settings, enabling timely interventions and strategic resource allocation. By offering personalized risk predictions, EWS significantly enhances patient outcomes and optimizesoverall care management in the adult population. This proactive and holistic approach, based on a diverse range of data, allows EWS to address the limitations posed by traditional health monitoring systems, providing a more accurate, adaptable, and comprehensive system aiding the physicians in making informed decisions promoting patient care.

[0013] However, the model does have some limitations. It is not recommended for pediatric populations. Its application might also need adjustments for other demographic zones, including the US, Europe, Middle East, and South East Asia. It is not suitable for use in emergency rooms for determining immediate triage categories or in place of existing triage algorithms. Results may vary for patients under non-invasive or invasive ventilator settings, those receiving IV inotropic support in critical care settings, pregnant women in labor, individuals with acute psychiatric conditions, and those with substance abuse. Additionally, the model does not recommend changes in critical care protocols or support end-of-life care protocols in clinical settings. Despite these limitations, the EWS offers a more accurate, adaptable, and comprehensive system that aids physicians in making informed decisions that promote patient care.Objectives of the invention:

[0014] The primary objective of this invention is to develop a system that utilizes a triphasic machine learning model to detect clinical deterioration accurately, compute risk scores, and predict mortality by analyzing a diverse range of patient data, including vital signs, comorbidities, clinical symptoms, and laboratory results.

[0015] Another objective of the present invention is to incorporate advanced machine learning techniques into the system to achieve a high accuracy rate ranging from 83% to 92%, providing a demographically relevant approach by primarily using Indian data.

[0016] Yet another objective of the present invention is to design the system to include a triphasic model that continually refines its predictive capabilities, offering a dynamic and adaptable solution in contrast to static models.

[0017] Yet another objective of the present invention is to create a system that provides clear risk categorization, enhancing its versatility and applicability across various healthcare environments and patient demographics.

[0018] Yet another objective of the present invention is to develop a system capable of efficiently processing large patient data sets, demonstrating scalability and adaptability.

[0019] Another objective of the present invention is to enhance the system's ability to integrate real-time, high-quality data from wearable devices, thereby providing a comprehensive and detailed view of a patient's health status.

[0020] The Further objective of the invention is to design the system to incorporate data from a wide array of sources, including vital signs, lab results, X-rays, and ECGs, into a comprehensive dashboard. This dashboard aims to provide a holistic view of a patient's health, assisting clinicians in making informed decisions.Summary of the invention:

[0021] The present disclosure proposes an adult early warning system (EWS) to predict risk score and deterioration of patients and method thereof. The following presents a simplified summary in order to provide a basic understanding of some aspects of the claimed subject matter. This summary is not an extensive overview. It is not intended to identify key / critical elements or to delineate the scope of the claimed subject matter. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.

[0022] In order to overcome the above deficiencies of the prior art, the present disclosure is to solve the technical problem to provide an adult early warning system for identifying a patient's clinical deterioration and predicting their risk score using a triphasic machine learning model.

[0023] The invention introduces an adult early warning system (EWS) designed to identify potential risks of patient clinical deterioration and mortality, enhancing patient careoutcomes. The adult early warning system comprises a server having a processor and a memory, which is connected to a database for efficient data management. The server is in communication with a user device via a network. The server communicates with a plurality of modules, which is configured to predict a risk of clinical deterioration and mortality of at least one patient. The plurality of modules are executed by the processor. The plurality of modules comprises client modules, an input module, processing module, prediction module, clinical pathway module, and risk prediction response and a patient overview dashboard module.

[0024] The client modules client modules comprise a data collection module, a standardization and pre-processing module, and an application programming interface (API) module.

[0025] In one embodiment, the data collection module is configured to collect raw data from various sources such as electronic medical records (EMRs), Data repositories and Databases. In one embodiment herein, the standardization and pre-processing module is configured to standardize and pre-process the collected data to ensure its integrity and compatibility with the system's processing requirements.

[0026] In one embodiment herein, the application programming interface (API) module is configured to transmit the pre-processed data from the data collection module and the standardization and pre-processing module to the server via the network.

[0027] In one embodiment, the input module is configured to receive the transmitted data from the server and initiate authentication and validation of input data for at least one patient, ensuring compliance with the required formats. In one embodiment, the processing module transforms and standardizes data from the input module to meet the necessary input criteria for subsequent analysis.

[0028] In one embodiment, the prediction module employs a triphasic model configured to analyze the data of at least one patient using a triphasic machine learning model. In Phase 1, an XGB model achieved 83% accuracy with vital signs and demographics. Phase 2incorporates a neural network, expanding inputs to vital signs, comorbidities, and symptoms, achieving 84% accuracy in predicting mortality and clinical deterioration. Phase 3 integrates an XGB model with comprehensive parameters, expanding to lab parameters and reaching 92% accuracy. Throughout all phases, the system provides numerical confidence levels and risk categorization, empowering caregivers with informed decision-making systems for patient care.

[0029] In one embodiment, the clinical pathway module is configured to provide a clinical decision support system for subsequent actions based on standard care protocols for at least one patient, based on the determined risk thresholds. In one embodiment, the risk prediction response and dashboard module offers a comprehensive overview, including predicted risk scores, contributing factors, and graphical representations of vital signs, lab results, X-rays, and ECG data. Additionally, the dashboard enables the generation of PDF reports for further analysis and documentation.

[0030] Furthermore, the system's database enables efficient data storage and retrieval, providing numerical confidence levels, risk categorization, and hourly risk score updates. The adult early warning system integrates advanced machine learning models with comprehensive data analysis and user-friendly interfaces, revolutionizing patient care and clinical decision-making processes.

[0031] In one embodiment, data from various sources is collected by the data collection module and standardized in the standardization and processing module. This standardized data is integrated into an API, which is sent through the input module via the network.

[0032] According to another aspect, the invention provides a method for predicting the risk of the patient's clinical deterioration using an adult early warning system. At one step, the data collection module collects the raw data obtained from various sources and the standardization and pre-processing module processes the raw data to ensure system's processing requirements.

[0033] At another step, the pre-processed data is transmitted to the server via the network through an API integration module and then transmitted to the input module for further validation via the server. At another step, the input module authenticates requests and validates input data as per the required format for at least one patient an Application Programming Interface (API) or via a web user interface on a user device. At another step, the processing module transforms and standardizes the data from the input module to meet the required input format for the model. At another step, the prediction module analyzes the health parameters of at least one patient using a triphasic machine learning model, which progressively enhances its accuracy through each phase. Initially, using an extreme Gradient Boost (XGB) model with vital signs data, it achieves 83% accuracy in predicting clinical deterioration.

[0034] The accuracy improves to 84% when the model expands its input to include vital signs, clinical comorbidities, and symptoms, employing a neural network to predict both clinical deterioration and mortality. In the final phase, by further incorporating lab parameters and reapplying the XGB model, the module reaches an accuracy of 92% for predicting clinical deterioration and mortality.

[0035] At another step, the clinical pathway module is configured to provide clinical pathways and standardized care protocols based on the output of the model. The final decision regarding recommending any test, procedure, or treatment remains at the physician's discretion. Further, at step, the risk prediction response and patient overview dashboard module is configured to display patient vitals, a 24-hour dial representing each hour, risk scores with color coding for different risk zones, and a graphical representation of contributing factors via the user interface. The dashboard provides a predicted risk score along with contributing factors and graphical representations for viewing vitals, lab results, X-rays, and ECG data.

[0036] Further, objects and advantages of the present invention will be apparent from a study of the following portion of the specification, the claims, and the attached drawingsDetailed description of drawings:

[0037] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate an embodiment of the invention, and, together with the description, explain the principles of the invention.

[0038] FIG. 1 illustrates a block diagram of an adult early warning system (EWS), in accordance to an exemplary embodiment of the invention.

[0039] FIG. 2A illustrates a snapshot representing the entry of vital signs of at least one patient, in accordance to an exemplary embodiment of the invention.

[0040] FIG. 2B illustrates a snapshot representing the entry of comorbidities of the at least one patient, in accordance to an exemplary embodiment of the invention.

[0041] FIGs. 2C-2D illustrate snapshots representing the entry of clinical symptoms of the at least one patient, in accordance to an exemplary embodiment of the invention.

[0042] FIGs. 3A-3B illustrate snapshots representing the obtained results of the at least one patient, in accordance to an exemplary embodiment of the invention.

[0043] FIGs. 4A-4B illustrate a snapshot representing an overall report of the at least one patient, in accordance to an exemplary embodiment of the invention.

[0044] FIG. 5 illustrates a flowchart of a method for identifying a patient clinical deterioration using an adult early warning system, in accordance to an exemplary embodiment of the invention.

[0045] FIG. 6 illustrates an architecture diagram of a method for developing and deploying adult early warning system (EWS) for predicting clinical deterioration and mortality, in accordance to an exemplary embodiment of the invention.Detailed invention disclosure:

[0046] Various embodiments of the present invention will be described in reference to the accompanying drawings. Wherever possible, same or similar reference numerals are used in the drawings and the description to refer to the same or like parts or steps.

[0047] The present disclosure has been made with a view towards solving the problem with the prior art described above, and it is an object of the present invention to provide an adult early warning system for identifying a patient's clinical deterioration and predicting the risk score of at least one patient using a Triphasic machine learning model.

[0048] According to one exemplary embodiment of the invention, FIG. 1 refers to a block diagram of an adult early warning system 100. The system uses an advanced machine learning model to achieve accuracy rates ranging from 83% to 92%. The system provides clear risk categorization and is adaptable to different healthcare settings and patient populations, offering versatility. The system is also scalable to handle large datasets of patient data.

[0049] The system 100 can be accessed through multiple methods, one of which involves integration with a user device using various approaches, including REST APIs. This integration method ensures seamless operation and generally employs HTTPS for secure communication. The user device is connected to a server 104, which includes a processor 106 and memory 108 for storing and executing instructions. The server 104 communicates with client devices via a network 102, enabling access to the system through web-based interfaces using secured credentials. Data is collected from various sources and then standardized and pre-processed in standardization and pre-processing module 124. The standardized data is integrated with the API integration module 126, facilitating seamless communication via the network 102.

[0050] In one embodiment, the term user device generally refers to any device configured to interact with the web-based system. This includes personal computers, cellular telephones, smartphones, personal data assistants (PDAs), laptop computers, tablet11computers, smartbooks, palm-top computers, wireless email receivers, multimedia internet- enabled cellular telephones, and similar personal electronic devices. These devices can connect to the network using wireless technologies such as Wi-Fi. In one embodiment, the network 102 could include, but is not limited to, Wi-Fi, Bluetooth, a wireless local area network (WLAN), an internet connection, and radio communication. The user device can be touchscreen or non-touchscreen and may operate on various operating systems, such as iOS, Windows, Android, Unix, Linux, and others.

[0051] In one embodiment, the system 100 is hosted on the server 104, which could be a computer or server designed for general or specialized purposes. This server could operate independently or as part of a broader configuration, which could include hardware servers, workstations, desktop PCs, laptops, tablets, mobile phones, mainframes, supercomputers, or server farms. Importantly, the server 104 is interconnected with the processor 106 and the memory 108, which is connected to a database 110 for efficient data management.

[0052] In one embodiment, the server 104 comprises plurality of modules, which is configured to predict a risk of clinical deterioration and mortality of at least one patient. The plurality of modules comprises client modules, an input module 112, a processing module 114, a prediction module 116, a clinical pathway module 118, and a risk prediction response and patient overview dashboard module 120. In one embodiment herein, the client modules are configured to perform multiple functions for identifying patient clinical deterioration. The client modules comprise a data collection module 122, a standardization and pre-processing module 124, and an application programming interface (API) module 126.

[0053] In one embodiment herein, the data collection module 122 is configured to collect raw data from various sources such as electronic medical records (EMRs), Data repositories and Databases. In one embodiment herein, the standardization and pre-processing module 124 is configured to standardize and pre-process the collected data to ensure its integrity and compatibility with the system's processing requirements.

[0054] In one embodiment herein, the application programming interface (API) module 126 is configured to transmit the pre-processed data from the data collection module 122 and the standardization and pre-processing module 124 to the server 104 via the network 102.

[0055] In one embodiment, the input module 112 is configured to receive the transmitted data from the server 104. The input module 112 authenticates and validates user data, including vital signs such as height, weight, BMI (auto-calculated), heart rate, systolic and diastolic blood pressure, respiration rate, temperature, oxygen saturation, and ventilation status.

[0056] The input module 112 also collects information on comorbidities and clinical symptoms, including histories of diabetes mellitus, hypertension, coronary artery disease, heart failure, arrhythmia, lung disease, central nervous system disorders, immunocompromised conditions, and chronic kidney disease, as well as current diagnosis and past surgical history. System-wise attributes cover lung health, central nervous system status, the type of ward the patient is in, and known allergies. Clinical symptoms include fever, cough, weakness, respiratory distress, cardiovascular issues, trauma, unconsciousness, and abdominal signs. Laboratory parameters encompass a complete blood count, basic biochemistry, coagulation profile, and inflammatory markers. In one embodiment, the processing module 114 processes the data from the input module 112, transforming and standardizing it to meet the required input format for the model.

[0057] In one embodiment, the prediction module 116 analyzes the health parameters of at least one patient using a triphasic machine learning model. This model progressively enhances accuracy through its phases. Using an Extreme Gradient Boost (XGB) model with vital signs data, it achieves 83% accuracy in predicting clinical deterioration. The accuracy improves to 84% when the model expands its input to include vital signs, clinical comorbidities, and symptoms, employing a neural network to predict both clinical deterioration and mortality. In the final phase, by further incorporating lab parameters and reapplying the XGB model, the accuracy improves to 92% for predicting clinical deterioration and mortality.

[0058] In one embodiment, the clinical pathway module 118 provides clinical pathways and standardized care protocols based on the output / risk threshold. In one embodiment, the Risk Prediction Response and Patient Overview Dashboard Module 120 is configured to display patient vital sign and lab parameter trends, a 24-hour dial, risk scores with color coding for different risk zones, and a graphical representation of contributing factors via the user interface. This dashboard can also generate a PDF. In one embodiment, the system 100 comprises a database 110, which is configured to enable communication with the system to store credentials and health parameters of various patients. The system 100 provides numerical confidence levels, risk categorization, and risk score data with an hourly dial system.

[0059] According to another exemplary embodiment of the invention, FIG. 2A refers to a snapshot 200 representing the entry of vital signs 208 of at least one patient. In one embodiment, the vital signs of at least one patient include height, weight, BMI (autocalculated), heart rate, systolic blood pressure, diastolic blood pressure, respiration rate, temperature, oxygen saturation, patient ventilation status, and past surgical history. Initially, the user enters the credentials and vital signs through the input module 112.

[0060] According to another exemplary embodiment of the invention, FIG. 2B refers to a snapshot 202 representing the entry of the current diagnosis and comorbidities 210 of at least one patient. In one embodiment, the comorbidities of at least one patient include a history of diabetes mellitus, hypertension, coronary artery disease, heart failure, arrhythmia, lung disease, central nervous system disorders, immunocompromised states, chronic kidney disease, and other clinical symptoms.

[0061] According to another exemplary embodiment of the invention, FIGs. 2C-2D refer to snapshots 204, 206 representing the entry of health attributes and clinical symptoms 210 of at least one patient. In one embodiment, the health attributes of at least one patient, such as attributes of lung health, are evaluated. This also includes other system evaluations like the Central Nervous System (CNS) assessment, the state of being immunocompromised, the presence of chronic kidney disease, the type of ward the patient is currently in, and any known allergies. The clinical symptoms 210 of at least one patient include fever, cough,weakness, respiratory distress, cardiovascular symptoms, trauma, unconsciousness, abdominal signs and symptoms, and any other symptoms. In one embodiment, the plurality of lab parameters includes a complete blood count, basic biochemistry, coagulation profile, and inflammatory markers.

[0062] According to another exemplary embodiment of the invention, FIGs. 3A-3B refer to snapshots 300, 302 representing the obtained results of at least one patient. In one embodiment, the snapshots 300, and 302 exhibit the user interface with vitals on the left side with the average of the last hour and a trend chart, a 24-hour dial that represents each hour, and a patient score with color coding for the different risk zones (default is <15 green (low risk), 15 - 30 - yellow (moderate risk), and >30 - red (high risk); below the dial - it represents the probability of clinical deterioration in the next 1 hr, 4 hrs, 8 hrs, 12 hrs, 24 hrs, and users can also view previous days' 24-hour dial, get a summary of the hour, and print the patient's current status.

[0063] The vitals of at least one patient, a 24-hour dial representing each hour, risk scores with color coding for one or more different risk zones, and a contributing risk factor plot of at least one patient are displayed via the user interface. According to another exemplary embodiment of the invention, FIGs. 4A-4B refer to snapshots 400, and 402 representing an overall report of at least one patient. The risk score categorization of at least one patient is represented in the overall report. The risk score of at least one patient is indicated in red, which represents the condition of the patient at high risk. The risk score is given based on the vital signs, comorbidities, and plurality of lab parameters of at least one patient.

[0064] According to another exemplary embodiment of the invention, FIG. 5 refers to a flowchart 500 of a method for identifying patient clinical deterioration using the adult early warning system 100. At step 502, the data collection module 122 collects the raw data obtained from various sources and the standardization and pre-processing module 124 processes the raw data to ensure system's processing requirements.

[0065] At step 504, the pre-processed data is transmitted to the server 104 via the network 102 through an API integration module 126 and then transmitted to the input module 112for further validation via the server 104. At step 506, the input module 112 authenticates requests and validates input data as per the required format for at least one patient through an Application Programming Interface (API) or via a web user interface on a user device. At step 508, the processing module 114 processes the data of at least one patient from the input module 112, transforming and standardizing it to meet the required input format for the model. At step 510, the prediction module 116 utilizes data from the processing module 114 and, through a triphasic machine learning model, achieves an accuracy range of 83% to 92%.

[0066] At step 512, based on the output from the prediction module 116, the clinical pathway module 118 provides clinical pathway and standardized care protocols for at least one patient through the user interface. Further, at step 514, the risk prediction response and patient overview dashboard module 120 provides a 24-hour dial of predicted risk scores, along with contributing factors and a graphical representation for viewing vitals, labs, X-ray, and ECG data and its components, and a PDF of the dashboard can be generated. The vitals of the patient, a 24-hour dial representing each hour, risk scores with color coding for different risk zones, and contributing risk factors are displayed via the user interface.

[0067] According to an exemplary embodiment of the invention, FIG. 6 illustrates the system architecture 600 for developing and deploying an artificial intelligence (Al)-based adult early warning system 100. This system architecture outlines the sequential phases integral to the system's development and deployment. In step 602, the process begins with the data sourcing phase, where information is collected from various sources, including but not limited to vitals and wearables data, comorbidities and symptoms, laboratory data from Electronic Medical Records (EMR) 614 and servers, doctors 618 and clinical knowledge bases 616.

[0068] In step 604, we move to the data ingestion phase. Here, data is collected and put into a centralized database 620. This data then traverses through a data pipeline 622 and is stored in a data repository 624 in structured formats. This phase is crucial for organizing the data and preparing it for further pre-processing stages. In step 606 includes the data pre-processing phase, which encompasses metadata management 626, ETL processes 628, and data transformation and harmonization 630. These processes improve data quality and consistency, setting the stage for subsequent analysis and model training.

[0069] In step 608, when it comes to data analysis, we utilize a variety of statistical systems 608. Propensity matching 632 unveils hidden patterns, while descriptive statistics 634 provide a summarized overview of the data. Correlation coefficients 636 establish relationships between various risk factors, and odds / hazard ratios 638 compare the possibility of outcomes. KM plots and survival charts 640 visualize time-to-event data, and visualization systems 642 simplify complex data representation. These systems, in concert, contribute to feature selection and model building of the prediabetes risk assessment system. In Step 610, we used a triphasic model to enhance accuracy at each stage. In Phase 1, an XGB model incorporating vital signs and demographics achieves 84% accuracy for predicting clinical deterioration. Phase 2 expands inputs to include comorbidities and symptoms using a neural network, maintaining 84% accuracy for both mortality and clinical deterioration.

[0070] The neural network model, built with the TensorFlow library, is specifically designed to predict the likelihood of these risks. It is trained with domain insights and data insights, and through various iterations, optimal performance is achieved. The model consists of four layers: the first layer has 128 neurons, the second layer has 64 neurons, the third layer has 32 neurons, and the final layer is a binary classification layer with one neuron. The 'ReLU' activation function is used in the first three layers, and the 'sigmoid' activation function is used in the last layer, tuning the model for binary classification. Phase 3 further integrates laboratory parameters with an XGB model, achieving 92% accuracy. Throughout all phases, the system provides numerical confidence levels and risk categorization, empowering caregivers with informed decision-making systems for patient care.

[0071] In step 612, the model is deployed for integration and usage through REST API 652 protocols. This architecture involves using an API management service 644 and an application service resource 650 for model inference. The model inference code is developed in the Python programming language 654. All resources are hosted on a virtualprivate network 648 securely. The API service 646 acts as the interface for the REST API, facilitating communication between the API management service and the web application. The API Management Service includes storage capabilities 658 for securely managing and storing model artifacts, data inputs, and results. In one such application of the integration, this model inference API is integrated into a web application code 656 to be deployed and used as a Web Application 660. This web application serves as a user-friendly platform for users or clinicians, enabling timely interventions.

[0072] Numerous advantages of the present disclosure may be apparent from the discussion above. In accordance with the present disclosure, an adult early warning system 100 is disclosed. The proposed invention provides a system for identifying patient clinical deterioration and predicting risk scores using at least one machine learning model. The adult early warning system 100 employs an advanced machine learning model to achieve high accuracy rates ranging from 83% to 92%. Using a triphasic model, it progressively refines predictions, offering a superior solution compared to static models.

[0073] This system 100 provides clear risk categorization and is adaptable to various healthcare settings and patient populations. Developed based on Indian data, the system 100 is scalable to handle large datasets, computationally efficient, and customizable to specific populations or settings. It offers explainability through contributing risk factors. Acting as a complementary system for clinicians, it enhances informed decision-making without replacing clinicians or diagnostic tests. The system 100 adheres to ISO 13485 standards, ensuring high quality and reliability. Functioning as a medical device (SaMD) and certified by ISO 13485, the system integrates algorithms for early detection and intervention of clinical deterioration, thereby enhancing patient outcomes and supporting clinicians.

[0074] It will readily be apparent that numerous modifications and alterations can be made to the processes described in the foregoing examples without departing from the principles underlying the invention, and all such modifications and alterations are intended to be embraced by this application.

Claims

5. CLAIMS: l / We Claim:

1. An adult early warning system (EWS) (100) for identifying patient clinical deterioration, comprising: a server (104) having a processor (106) and a memory (108), which is connected to a database (110) for efficient data management, wherein the server (104) is in communication with a user device via a network (102), the server (104) comprises plurality of modules, which is configured to predict a risk of clinical deterioration and mortality of at least one patient, wherein the plurality of modules comprises: client modules, which are configured to perform multiple functions for identifying patient clinical deterioration, wherein the client modules comprises: a data collection module (122) configured to collect raw data from various sources such as electronic medical records (EMRs), Data repositories and Databases through a data collection module 122; a standardization and pre-processing module (124) configured to standardize and pre-process the collected data to ensure its integrity and compatibility with the system's processing requirements; and an application programming interface (API) module (126) configured to transmit the pre-processed data from the data collection module (122) and the standardization and pre-processing module (124) to the server (104) via the network (102); an input module (112) configured to receive the transmitted data from the server (104) to authenticate and validate the user's data, including vitals, comorbidities, and laboratory parameters; a processing module (114) configured to transform and standardize the data received from the input module (112) to fit the required input format for the model;a prediction module (116) equipped with a triphasic model configured to process the standardized health parameter data from the processing module (114) and generates predictions with an accuracy ranging from 83 to 92%, based on these predictions, the clinical pathway module (118) provides standardized care protocols and clinical pathways; a risk prediction response and patient overview dashboard module (120) configured to provide a predicted risk score along with contributing factors, wherein the risk prediction response and patient overview dashboard module (120) is configured to provide graphical representations for viewing vitals, labs, X-rays, and ECG data for each component, and generate a PDF of the dashboard, display vitals of the patient, a 24-hour dial representing each hour, risk scores with color coding for different risk zones, and contributing risk factors via a user interface, thereby acting as a complementary system for clinicians, it enhances informed decision-making without replacing clinicians or diagnostic tests, and the adult early warning system (EWS) (100) adheres to ISO standards, ensuring patient safety and reliability, thereby functioning as a medical device (SaMD) and being certified by ISO 13485, the adult early warning system (EWS) (100) integrates algorithms for early detection and intervention of clinical deterioration of the at least one patient, thereby enhancing patient outcomes and supporting clinicians, whereby the adult early warning system (100) provides a holistic approach for detecting early signs of clinical deterioration of the at least one patient.

2. The adult early warning system (100) as claimed in claim 1, wherein the health parameters of the at least one patient include vital signs, comorbidities, clinical symptoms, and plurality of lab parameters.

3. The adult early warning system (100) as claimed in claim 1, wherein the vital signs of the at least one patient include heart rate, systolic blood pressure, diastolic blood pressure, rate of respiration, temperature, oxygen saturation, and other demographical data.

4. The adult early warning system (100) as claimed in claim 1,wherein the comorbidities and clinical symptoms of the at least one patient include a history of diabetes mellitus, hypertension, coronary artery disease, heart failure, arrhythmia, lung disease, central nervous system disorders, immunocompromised conditions, and chronic kidney disease, as well as current diagnosis and past surgical history, and the system-wise attributes cover lung health, central nervous system status, the type of ward the patient is in, and known allergies, clinical symptoms include fever, cough, weakness, respiratory distress, cardiovascular issues, trauma, unconsciousness, and abdominal signs.

5. The adult early warning system (100) as claimed in claim 1, wherein the plurality of the lab parameters includes a complete blood count, basic biochemistry, coagulation profile, and inflammatory markers.

6. The adult early warning system (100) as claimed in claim 1, wherein the adult early warning system (100) is configured to facilitate a complementary and adjunct system for clinicians, it enhances informed decision-making without replacing clinicians or diagnostic tests, adhere to ISO standards, thereby ensuring high quality and reliability and functioning as a medical device (SaMD) and certified by ISO 13485, and integrate algorithms for early detection and intervention of clinical deterioration, thereby enhancing patient outcomes and supporting clinicians.

7. The adult early warning system (100) as claimed in claim 1, wherein the adult early warning system (100) comprises the database (110), which is configured to enable communication with the system to store credentials and health parameters of various patients.

8. The adult early warning system (100) as claimed in claim 1, wherein the clinical Pathway module (118), based on the output, provides clinical pathway and standardized care protocols.

9. The adult early warning system (100) as claimed in claim 1, wherein the adult early warning system (100) provides numerical confidence levels, risk categorization, and risk score data with an hourly dial system.

10. A method for predicting the risk of the patient's clinical deterioration using an adult early warning system (100), comprising: collecting, by a data collection module (122) the raw data obtained from various sources, and processing the raw data by a standardization and pre-processing module (124) to ensure system's processing requirements; transmitting, the pre-processed data to a server (104) via a network (102) through an API integration module (126), and then transmitted to an input module (112) for further validation via the server (104); enabling, by the input module (112), authentication of requests and validation of input data as per the required format of at least one patient through an Application Programming Interface (API) or via a web user interface on a user device; analyzing, by a processing module (114), data from at least one patient from the input module (112) processed, which is transformed and standardized to meet the required input format for the model; integrating, by a prediction module (116) data from the processing module (114) and providing input with accuracy ranging from 83% to 92%; alerting, a clinical pathway module (118) provides clinical pathways and standardized care protocols for at least one patient through the user interface based on the output from the prediction module (116); and displaying, by a risk prediction response and patient overview dashboard module (120), a predicted risk score along with contributing factors and providing graphical representations for viewing vitals, labs, X-ray, and ECG data for each component, and generating a PDF of the dashboard, and displays vitals of the patient, a 24-hour dial representing each hour, risk scores with color coding for different risk zones, and contributing risk factors via the user interface.

Images