Engineer, AI, IoT, digital transformation, strategy, healthcare innovation, preparedness, researcher, author
AI technologies and more specifically Large Language Models (LLMs) like ChatGPT are the divas of clinical tech talk. While important, they are only part of a broad spectrum of emerging technologies poised to transform clinical care and doctor-patient relationships. These include #wearables and patient #monitors based on advanced physiological sensors that are similarly changing healthcare and fitness landscapes by synergistically exploiting AI to advance physician-centered patient specific care. Our research and use-case investigations suggest that advanced monitors, including wearables, can help counter AI intrusions into doctor-patient relationships. Specifically, monitoring augments subjective patient symptoms with quantitative physiological data that arm doctors with near realtime measurements, trends, and contextual insights. They also help patients by making symptoms and asymptomatic conditions easier to understand and visualize. Patients in this paradigm become collaborating partners instead passive passengers on the healthcare-AI express. What about AI? The best monitoring systems have been using AI technologies including machine learning and deep learning to improve measurement accuracy, precision, and quality. LLMs like ChatGPT performed best in our evaluations by exploiting post-training monitoring data and research to produce more up to date and insightful responses. We frequently use Retrieval Augmented Generation (RAG) techniques to make ChatGPT aware and direct it to use data and information produced after its last training corpus update. RAG is indispensable in many LLM based applications, so I’ve included a link to a video presenation by IBM research scientist Marina Danilevsky. In summary, continuous and episodic physiological monitoring based on advanced #sensors are poised to transform clinical care by arming doctors, patients, and LLMs with unprecedented data and information. We noted in our investigations that ChatGPT produced more accurate and insightful answers when it had access to baseline and near realtime monitoring data sets and up to date research. Our investigations and use case evaluations suggest that monitoring technologies and data centered strategies can improve ChatGPT performance while empowering clinicians and strengthening doctor-patient collaboration and trust. Overall, monitoring technologies currently empower clinicians and doctor-patient collaboration more than LLMs like ChatGPT. This is an evolving field of research driven by continuous technological innovations so stay tuned and reach out if you have questions. https://lnkd.in/ge-MQBeU #llm #chatgpt #rag #retrievalaugmentedgeneration #healthcareinnovation #healthcareai #wearablemonitors #remotepatientmonitoring #doctorpatientrelationship
CIO / Senior Level Information Technology Professional, EMBA
2moLiked the video but not sure of RAG. Not comfortable with augmenting the response from “a” data source. With business entities controlling these data sources, it’s a Pandora’s box of reliability and bias. LLM’s that do retrieval augmentation not just from a pre-selected data source but multiple sources would probably make sense but then it’s a ranking challenge- which source is more reliable and unbiased!