Transforming Diabetes Care with AI: Insights from Healthcare Experts

Artificial intelligence/Machine learning (AI/ML) is revolutionising every sphere of healthcare, including diabetes management. Since India has a huge burden of diabetes cases, harnessing the power of artificial intelligence in diabetes treatment can successfully devise tailor-made solutions for controlling this medical condition. 

To dissect and examine the potential of AI in diabetes management, Fitterfly hosted an informative and insightful panel discussion led by Dr. Arbinder Singal (Co-founder & CEO, Fitterfly Healthtech) and Dr. Om J Lakhani (Consultant Endocrinologist, Zydus Hospital and Co-founder, Mellitus Healthtech). Dr. Neelesh Bhandari, (Vice President, Physician Partnerships, Fitterfly), moderated the session. In this write-up, we explore the focal points of their analysis on the impact of artificial intelligence in diabetes care, in case you have missed the event.

Current Role of AI in Diabetology

Dr. Arbinder Singal noted that technology is emerging as a key player in every stakeholder’s life in the healthcare value chain. Simple cameras in smartphones can seamlessly detect diabetic retinopathy. AI in diabetes management can also predict the risk of developing complications by analysing the vast pool of genomic, behavioural, lifestyle, and sleep data. The large-scale multifactorial, multidevice data can transform diabetes management by improving medical adherence and modifying lifestyles among patients.

Fig1: Dr Arbinder Singal [MBBS (Gold Medalist) M.Ch (AIIMS, Gold Medalist), DNB, MNAMS, Stocker Fellow (OSU, USA)]

Dr. Om J Lakhani also resonated with Dr. Arbinder Singal’s views on the revolutionary role of technology in healthcare. According to a recent study by Jaycee M. Kaufman et al, a smartphone app could predict the diagnosis of diabetes simply by analysing patients’ voice samples. Here, ML tools can analyse data to the granular level and determine whether a person is living with diabetes. In the future, there is a possibility that physicians will diagnose medical conditions among patients simply by scanning patients via smart gadgets. Most physicians believe that despite doomsday prophecies, AI is the future of healthcare, and it will pave the way for personalised therapies.

Implementation of AI-driven Chatbots for Enhancing Patient Engagement

Many clinicians are now implementing AI-driven chatbots to simplify the treatment process. Dr. Om Lakhani generously shared a few of his ongoing AI tools that can assist doctors in complex diagnoses of endocrinal disorders, creating prescriptions, and organising patients’ reports in a systematic manner. He and his team have developed a tool named Diabetology.co.in, where physicians can input patients’ anthropometric data and vital endocrine parameters like the HbA1c, fasting blood sugar, and creatinine to generate an individualised prescription. 

Fig 2: Dr. Om J Lakhani [ MBBS, MD (Govt. Medical College, Baroda, Gold Medalist), DNB (Endocrinology, Gold Medalist), Specialty Certificate in Endocrinology (UK)]

Another AI-driven chatbot named EndoAI.co.in helps doctors in making complex medical diagnoses. The team has collaborated with Endotext to create a huge database of Endocrinology-based scientific papers. Doctors can input their concerns in any aspect of Endocrinology to gain scientific, clinically validated answers. Despite the functioning being similar to ChatGPT, it is enriched with in-depth analysis of research papers and developed exclusively for healthcare professionals.

How is Fitterfly Implementing AI in CGM?

Dr. Arbinder Singal revealed that Fitterfly is enriched with data from over 10000 members who have undergone CGM analysis in our Diabetes Programs. The DTx company also has a food database consisting of 32 cuisines and 37000 recipes. We combined the data from all these sources into our AI tool named JEDi. It is an empathetic personalised AI coach guiding members in determining the nature of their food (healthy/unhealthy), providing metabolic analysis of meals, answering FAQs on diabetes in a person-centric manner, and suggesting curated diet plans. 

Fitterfly is also working on adding recipe videos for customised diets and integrating the chatbot with users’ grocery lists for seamless shopping of ingredients. Fitterfly is also on the verge of launching a complete AI nutrition coach. More than 60 dietitians, 20 physiotherapists, and 20 psychologists have created over 2.2 million data points for the functioning of this AI tool.

Development of Virtual CGM Model

One of the major drawbacks of CGM devices is the high expenditure (around Rs 5000 per device) associated with them. To solve this problem, Fitterfly is working towards developing a virtual CGM (vCGM) based on the data of over 10000 members and the nutritional composition of 37000 Indian foods. Studies revealed that the predicted values of virtual CGM are nearly 85 to 90% accurate compared to actual CGM device values. Fitterfly has also filed a patent for this technology. This AI tool enables members to make healthy food choices.

Limitations of CGM Devices

The actual CGM devices fail to demonstrate accurate data in 10-12% readings. There have been reports of false positive hypoglycaemia and abnormal fluctuations in blood sugar values. Proper calibrations of CGM devices can check these limitations. Fitterfly has built an AI-driven model to predict blood sugar levels at different times of the day based on their HbA1c level. If the readings of the actual CGM device and the virtual CGM model do not match, it triggers an auto-driven calibration logic which solves the accuracy limitation to a large extent.

Efficacy of AI-powered Insulin Pumps

AI significantly contributed to the development and functioning of cutting-edge insulin pumps. Currently, many endocrinologists prescribe hybrid closed-loop insulin pumps based on proprietary machine-learning algorithms, especially for managing Type 1 diabetes. They lead to improved glycaemic control and reduce incidences of diabetic ketoacidosis and hypoglycaemia. While the hybrid pumps can correctly interpret the dosage of basal insulin they cannot do the same for bolus insulin. Research on complete close-loop AI-powered pump development for managing both insulin dosages is ongoing. Scientists are also developing gadgets that could titrate both insulin and glucagon dosage.

 AI-based Data Mining: The Future of Diabetes Research

Data mining is considered the future of diabetes research as it can radically transform clinical trials. It can accelerate clinical trials manifold through real-time data collection, adjustments, analysis, and automated paper publication with correct referencing. The entire drug discovery process to drug trials will be fast-forwarded by implementing AI-based data mining.

Artificial intelligence and machine learning are rapidly reshaping the landscape of diabetes management. With India’s substantial diabetes burden, leveraging the potential of AI/ML offers tailored solutions for effective control. The insights shared by Dr. Arbinder Singal and Dr. Om J Lakhani shed light on how technology is revolutionising healthcare and the limitless possibilities it presents. We invite you to delve deeper into this transformative conversation by watching the full discussion in the Transforming Diabetes Care with Artificial Intelligence: Clinical Insights video below. 

About Us

Fitterfly is on a mission to fuel the global metabolic health revolution through the power of Digital Therapeutics (DTx). We partner with physicians to deliver advanced customised digital therapeutics solutions aimed at enhancing the health outcomes of their patients.

By partnering with Fitterfly, you can save time in your clinics, ensure continuity of care for your patients, get research publications, and also enhance your revenue. To know more, please fill out the form below or call us at 080-47093933.