🚨 Pulse of Innovation 🚨

Breaking news in the healthcare AI

Faster Drug Discovery and Better Matching of Patients with the Drugs: The AI at work

• Two recent articles published in two different Nature.com journals, one on machine learning powered detection of potent drug compound for Parkinson’s disease and the other on AI powered matching of cancer patients to the best drug therapy they are likely to respond to.

• The Cambridge University researchers used machine learning approach to identify small molecule inhibitors for α-synuclein aggregation, a process implicated in the Parkinson’s disease, for which there is currently no approved disease modifying drugs. With implementation of the machine learning based structural pathways the identification of the compounds was faster (10 times) and significantly cheaper and more over the discovery provided more potent compounds overcoming an initial time consuming road block.

  

   

• The NIH Cancer Institute researchers have developed PERCEPTION (PERsonalized Single-Cell Expression-Based Planning for Treatments In ONcology), an AI powered precision oncology computational model which uses matched bulk and single-cell (sc) expression profiles from large-scale cell-line drug screens to build treatment response models. PERCEPTION showed success in predicting responses to targeted therapies in cultured and patient-tumor-derived primary cells, and in two clinical trials for multiple myeloma and breast cancer.

  

🧑🏼‍🔬 Bench to Bedside👨🏽‍🔬

Developments in healthcare AI research and innovations

Knowledge of GPT-4 is getting better in Ophthalmology, inching closer to experts

• A recent article published in the PLOS Digital health performed a cross sectional study of Large Language Models (LLM) to ophthalmology expert to determine the knowledge and reasoning displayed by the models.

• The GPT-3.5, GPT-4, PaLM 2, LLaMA, expert ophthalmologists, and doctors in training were trialed on a mock examination of 87 questions. Performance was analyzed by masked ophthalmologists and they graded the accuracy, relevance, and overall preference of GPT-3.5 and GPT-4 responses to the same questions..

• The performance of GPT-4 (69%) was superior to GPT-3.5 (48%), LLaMA (32%), and PaLM 2 (56%). GPT-4 compared favorably with expert ophthalmologists (median 76%, range 64–90%), ophthalmology trainees (median 59%, range 57–63%), and unspecialized junior doctors (median 43%, range 41–44%).

👩‍⚕️AI in the Clinic 🏥

Real-world and practical use of healthcare AI in clinic

AI chatbot for post-partum health needs: Path to equitable access to care resources

• The Penn designed a chatbot Penny - an AI enabled technology that has more than 70% accuracy matching the answers of healthcare providers for the “fourth trimester” post-partum care needs. This is an interactive and hybrid platform that has AI enabled chats as well as text messaging service which will alert the Penn team on certain occasion and a human will take over and answer all the health concerns.

• This technology had engagement rate of 92% and non-Hispanic black women were more likely to use and recommend it to others compared to white patients. This tool can screen for post-partum depression and hypertension effectively which are often under or delayed diagnosed.

🩺 Start-Up Stethoscope 💵 

Trending startups and technologies impacting clinical practice

Optimizing in-patient workflows for cost-savings using AI: Kontakt.io

• Kontakt.io which specializes in real-time data capturing through AI powered cloud and internet of things (IoT), recently raised $47 million in series C funding from the Goldman Sachs.

• The company’s platform can capture patients at different levels of care during their in-patient stay and the AI can better detect the care delivery process and pathways. This better understanding of care delivery pathways can better predict resource utilization, staffing needs and reduce wastage of resources by improved prediction of care delivery model.

  

🤖Patients First, AI Second🤖

Ethical and regulatory landscape of healthcare AI

National Academy of Medicine Code of Conduct framework for AI in Healthcare

• The National Academy of Medicine (NAM) has established a steering committee to appropriate and draft a Code of Conduct for use of AI in Healthcare part which they released a framework on drafting of the code considering the rapidly changing healthcare AI field. Their commentary shows core principles and areas of focus in understanding and establishing the code of conduct.

• The core principles identified are engaged, safe, effective, equitable, efficient, accessible, transparent, accountable, secure and adaptive use of AI in healthcare. In addition they have described commitments that the code of conduct shall provide that is applicable implementation of AI in healthcare