Artificial Intelligence
The interim chief digital and information officer and CMIO at New York's Mount Sinai Health System raises the curtain on the artificial intelligence the organization is using today – and what tomorrow is likely to look like.
The genetic testing company used GPT-4o to create an artificial intelligence-driven tool that helps doctors create screening plans based on patient data, including personal risk factors and family history.
Although executives may feel pressured to use "error-prone" generative AI, mature, proven machine learning applications should be deployed in clinical settings first, contends Ran Balicer, HIMSS board of directors member.
Why wait 10 days to see a dermatologist when an AI app with nearly the same accuracy as a clinician can treat you the same day? That's a question providers may soon have to answer, says Mount Sinai's interim chief digital and information officer.
Tx-LLM was fine-tuned from Google's Med-PaLM 2 and created to analyze a variety of chemical or biological entities to assist with the drug-discovery pipeline.
In some ways, AI could achieve the same accuracy as physicians sooner rather than later. And some aspects of clinical care may migrate away from physicians, says Mount Sinai's interim chief digital and information officer.
"Healthcare organizations cannot rely on legacy technologies to detect and respond to today’s attacks," says Ricardo Villadiego, CEO of cybersecurity firm Lumu.
The Wisconsin medical center's HIMS department reported a complete transformation in their workflow: It could save up to 40% of time when abstracting a large volume chart, which equates to 16.7 hours per week.
Collaborations and integrations aim to streamline access to patient medical histories via national networks, drive more proactive pediatric care, expand behavioral telehealth and pioneer more hospital automation.
Generative AI and large language models sometimes give inaccurate answers, says Dr. Calum Yacoubian, director of NLP healthcare strategy at IQVIA. He describes techniques that can improve algorithms' accuracy.