Machine Learning based clinical decision support system for early diagnosis from real time physiological data

Recent advancements in science and technology, in every field, made it possible to develop systems which are/can monitor and store our actions either in the form of electronic or manual data. Historical data has been used to predict actions by analyzing relevant information from it. Analysis and rel

Project Title

Machine Learning based clinical decision support system for early diagnosis from real time physiological data

Project Area of Specialization

Artificial Intelligence

Project Summary

Recent advancements in science and technology, in every field, made it possible to develop systems which are/can monitor and store our actions either in the form of electronic or manual data. Historical data has been used to predict actions by analyzing relevant information from it. Analysis and reliable identification of our actions are possible using Machine Learning algorithms ranging from simple classification to clustering problems.

In last two decades, the development of patient monitoring system has been significantly increased especially in the area of general medicine, vital sign monitoring, Clinical Decision Support Systems (DSS), smart alarm monitoring and other computer aided diagnostic systems (Skythberg et. al 2016). Currently, clinical decision support systems and expert systems are considered as the most common areas used by clinicians to make better decisions to improve healthcare. With the usage of new technological tools (such as ML) and advanced sensors for physiological parameter registrations, new prospects are offered to researchers and scientists in healthcare. Consequently, healthcare professionals ultimately need new electronic systems (improved DSS) processing clinical data for early diagnosis of diseases.

Pakistan with a large population living in rural areas and with limited access to modern facilities; such as education, transport, telecommunication, justice, security and especially in healthcare. There is a need to develop systems which can monitor/analyze and predict symptoms/disease and maintain a subject’s history without having physical access to hospitals. The presented work is proposed to develop a prototype decision support system application for early disease diagnosis using physiological to facilitate people living in backward areas or having no direct access to hospitals.

Project Objectives

To design and develop a prototype of a clinical decision support system for early diagnosis of chronic diseases using real-time physiological data. The proposed objectives of the project are:

• Analyze performance of different machine learning supervised algorithms which have reasonable accuracy and are computationally less expensive.
• Design a framework of Clinical Decision Support System using Rule-based expert systems and associated software components;
• Knowledge base, Database, Inference Engine, User Interface etc.
• Implement proposed design and test against standard dataset.

Project Implementation Method

The proposed project implementation involves a scalable, structured and phased approach consisting of pre-defined inputs, activities and outputs which deliver a solution that will meet project objectives. The methodology is divided into phases and each phase its own identity and significance, the respective phases:

1. Initiate Phase

In this phase, the project group members plan out the project activities, resources and timelines. The subsequent phases of the project are built on the foundation created during this phase. The list of activities carried out during this phase are:

• Meetings with supervisor and Problem understanding/review
• Project scope, goals and objectives
• Literature review
• Selection of wearable sensors for dataset collection
• Project target milestones (key deliverables)
• Project plan outline

1. Design and Development Phase

In the Design phase, the objectives and needs in detail were explored and started architecting the solution that will best meet the project parameters. The key activities of this phase are:

• Dataset collection and analysis of dataset
• Applying relevant machine learning algorithms for data analysis
• Design and modelling of inference engine
• Design of decision support system using Rule-based expert system
• Design of software framework of components; Knowledge Base, Database, User Interface etc.
• Development of software interfaces/components of proposed solution
• Stand-alone/Independent testing of designed software components

1. Implementation and Integration Phase (may include design modification)

In this phase, the configuration and solution building are performed based on the project design. This phase consists of the following activities:

• Implementation of pilot software solution
• Implementation of test application
• Integration of software components using custom-designed API

1. Testing Phase

The final phase is Testing, which include activities such as:

• Test script creation of individual software unit testing
• End-to-end testing of data flow, verification and validation of reliable working of application interfaces
• Modification/upgradation of software components/libraries

Project solution readiness testing

Benefits of the Project

Different techniques and methods are used to improve the provision of healthcare services. The software solution presented a real-time interactive e-healthcare system to help users/patients monitor healthcare data; 5 wearable sensors; BP, Temperature, Hear rate, Oximeter, ECG and 3 ambient sensors: Light, Temperature, Humidity. All sensors data (dynamic in nature) has been handled intelligently and stored in a scalable database. Furthermore, Bayesian inference model classifies the health status (chronic diseases) of the users/patients. Such preliminary health status helps users/patients to assess health while staying at home. The user will be able to interact with the decision support system using textual input system for real-time disease diagnosis. In case of any anomaly within data, the user/patient can communicate with a doctor. Hence, the prime benefit of the proposed solution is to have a real-time assessment of health-related data and ability to connect users with the doctors to have real-time consultation or guidance, prior to visiting a hospital in person.

Moreover, the designed solution is to facilitate users/patients in isolated/remote communities by enabling them to collect physiological health-related sensors data at their homes, and the developed system will transport collected data intelligently to doctors/specialists far away without having to travel to visit them.  Furthermore, the project solution will one step closer towards enhancing the quality of life and well-being for the remote living people.

Technical Details of Final Deliverable

The technical details of the proposed project proto-type consist of the following:

Working Procedure:

• Literature Review
• Data collection and Attribute
• Proposed design and Component
• Implementation
• Integration and Testing
• Report Writing

Type of industry:

• Medical Industry
• Health Industry

Technologies:

• Artificial Intelligence
• Expert Systems

Final Deliverable of the Project

Software System

Core Industry

Medical

Other Industries

Others , Health

Core Technology

Artificial Intelligence(AI)

Other Technologies

Internet of Things (IoT), Others

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com