TensorHealth-NewsLetter on Medical AI

Exploring the Intersection of Artificial Intelligence and Healthcare Innovation

Opinion AI

What else could weight-loss drugs treat?

Exploring the Potential of Weight-Loss Drugs Beyond Obesity Management
Once the sci­ence is fully under­stood, products such as Ozempic, Wegovy and Moun­jaro could poten­tially be used to com­bat addic­tion, car­di­ovas­cu­lar dis­ease, Alzheimer’s and other ill­nesses.

News bits and bytes:

Foundation model of ECG diagnosis:
ECG interpretation remains a challenging task that requires extensive training (usually more than 12 years), is time-consuming and error-prone even for experienced cardiologists, and is especially difficult to deliver in remote and underserved regions due to the scarcity of specialists and limited resources

Scientists Aim to Create AI-Powered Virtual Human Cell
The virtual cell could aid in studying cancer, predicting virus impacts, and enabling personalized medicine by simulating treatments on digital patient models.

ADHD in adults: Atomoxetine and stimulants are best for managing symptoms, study reports

The selective noradrenaline reuptake inhibitor atomoxetine and stimulants (amphetamines and methylphenidate) are the “most effective treatments” for managing the symptoms of attention deficit/hyperactivity disorder (ADHD) in adults, researchers found

Microsoft's AI Health Division:
Microsoft’s arti­fi­cial intel­li­gence head Mustafa Suley­man is build­ing a team focused on con­sumer health by hir­ing staff from Google Deep­Mind. Health has become one of the growth areas in the AI boom. Con­sumers have often turned to the web for heal­th related quer­ies, and a Deloitte sur­vey this year found that 48 per cent of respond­ents asked gen­er­at­ive AI chat­bots such as Chat­GPT, Gem­ini, Copi­lot and Claude health ques­tions. (source: financial times)

The Rise Of Reasoning Models: AI Is Becoming More Human-Like
Artificial intelligence (AI) has witnessed remarkable advancements over the past few years, particularly in the development of reasoning models. These models are designed to emulate human-like reasoning capabilities, allowing them to solve complex problems across various domains. A significant factor contributing to the recent surge in reasoning models is the availability and utilization of diverse reasoning datasets.

Prediction of metabolic subphenotypes of type 2 diabetes via continuous glucose monitoring and machine learning:
The current classification system for type 2 diabetes and prediabetes doesn't account for the various ways glucose regulation can malfunction. Research demonstrated that prediabetes has distinct metabolic subphenotypes that can be identified by analyzing glucose curve patterns during at-home oral glucose-tolerance tests using continuous glucose monitors. In a study of 32 individuals with early glucose dysregulation, researchers identified dominant or co-dominant subphenotypes, with 34% showing muscle or hepatic insulin-resistance phenotypes and 40% displaying β-cell-dysfunction or impaired-incretin-action phenotypes. Machine learning models trained on glucose time series data showed high accuracy in predicting these subphenotypes, achieving AUCs of 95% for muscle insulin resistance, 89% for β-cell deficiency, and 88% for impaired incretin action. The ability to identify these metabolic subphenotypes through at-home continuous glucose monitoring could improve risk assessment for individuals with early glucose dysregulation.