AI Empowered Healthcare: From Prevention To Care

Topic Name:

EPITOME – EPIlepsy and Tumor detection through Optimised Machine learning Engine

Visual Abstract Image

Objectives

• Classify Multigrade Brian Tumor (Glioma and medulloblastoma) using Radiogenomics and create a predictive model that can forecast the likelihood of patient survival for a specified period using radiomic features.
• Develop AI-powered diagnostic tool for histopathology that can assist in grading tumor types, monitoring disease progression, quantify biomarkers and predict therapeutic efficacy
• Design and develop a wearable device prototype to capture electrical signals to detect the onset of epilepsy seizures in real time, providing timely alerts and warning to patients, care givers and health care professionals

Issues Involved

Data Quality and Variability: Limited availability of high-quality and consistent datasets across institutions, affecting tumor classification, seizure detection, and histopathology AI models. Variations in preparation methods, staining, or imaging contribute to inconsistencies.

Model Generalizability: Difficulty in creating AI models that work accurately across diverse patient populations and institutions due to data-shift, inter-observer variability, and individual differences in seizure patterns or tumor presentations.

Clinical Validation and Regulatory Approvals: Lack of robust, long-term clinical validation for AI tools and wearable devices, leading to delays in FDA or similar approvals and limited integration into healthcare workflows.

Real-Time Functionality: AI tools and wearable devices often require substantial processing time and stable internet connectivity, making them challenging to use in scenarios requiring rapid clinical decisions.

Device Usability and Patient Comfort: Wearable devices must be lightweight, comfortable, water-resistant, and equipped with long battery life for continuous monitoring, which poses significant design challenges.

Cost and Accessibility: High costs of AI tools and wearable devices limit accessibility, especially in low- to middle-income regions, underscoring the need for cost-effective solutions.

Rare Condition Detection and Ethical Concerns: AI systems may struggle to identify rare or complex pathologies and must address ethical concerns like false alarms, caregiver response, and misuse of sensitive healthcare data.

Team Lead

Name: Dr. Hima Deepthi Vankayalapati

Email ID: himadeepthi.vankayalapati@nmims.edu

Team Members

Name: Dr. Vaishali Kulkarni

Email ID: Vaishali.Kulkarni@nmims.edu

Name: Dr. Radhika Chapaneri

Email ID: Radhika.Chapaneri@nmims.edu

Name: Dr. Ami Munshi

Email ID: Ami.Munshi@nmims.edu

Name: Dr. Kalyani Barve

Email ID: Kalyani.Barve@nmims.edu

Name: Dr. Ravi Chandrashekhar Reddy

Email ID: ravichandra.danduga@nmims.edu