BREAKTHROUGH: AI Identifies 93.8% of Patients with Adverse Social Determinants of Health!!
In a groundbreaking development, researchers discovered that artificial intelligence (AI) models can effectively identify patients with adverse social determinants of health (SDoH) within electronic health records. Forget searching for needles in haystacks, this study showed that AI identified a whopping 93.8% of patients with these adverse health determinants, compared to only 2% identified by traditional ICD-10 codes.
“This is not just about algorithms acing medical exams, people! These findings could be a game-changer in revolutionizing how we tackle health disparities,” said Dr. Danielle Bitterman, a faculty member in the Artificial Intelligence in Medicine Program at Mass General Brigham.
The researchers trained AI models to identify any mentions of six SDoH – employment, transportation, social support, housing, relationships, and parental status – in electronic health records of cancer patients. Results showed an astounding 93.8% identification rate for adverse SDoH by AI models, compared to a paltry 2% using ICD-10 codes.
Implications for primary care providers (PCPs)? Dr. Bitterman said that validated technologies like these could help PCPs to review patient records proactively and identify those who have adverse SDoH. This could lead to multidisciplinary care planning and ultimately reduce clinician burnout.
Future research efforts are already underway to evaluate how AI-assisted SDoH screenings can translate into real clinical settings. The chief research goal is to understand whether large language model-assisted SDoH screening improves patient care and outcomes.
Anything else to add? Well, not only did the research use large language models to detect SDoH but it also leveraged generative AI to create synthetic clinical data for fine-tuning better performing, smaller AI models. Early days, but promising nonetheless, and this could be a trailblazing avenue for future research that could help preserve patient privacy.
What do you think about these findings? Could AI revolutionize healthcare as we know it? Let us know your thoughts and comment below!
IntelliPrompt curated this article: Read the full story at the original source by clicking here