## Predicting Social Determinants of Health via AI A recent study focuses on using artificial intelligence to identify social determinants of health (SDoH) from clinical notes. Social determinants of health, such as economic and social conditions, significantly influence patient outcomes but are often not included in structured data. This study explores the use of reasoning models and large language models (LLMs) to classify ICD-9 codes related to SDoH in hospital admission data, using the MIMIC-III dataset. ## Methodology and Results The researchers leveraged existing ICD-9 codes for prediction on admissions, achieving an F1 score of 89%. The study also identified missing SDoH codes in 139 admissions. The results suggest that the integration of AI models can significantly improve the accuracy and completeness of data related to social determinants of health, leading to a better understanding of patient needs and more targeted healthcare interventions. The research also provides the code to reproduce the results, promoting transparency and reproducibility in AI research applied to healthcare. ## Background The analysis of social determinants of health is crucial for addressing health inequalities and improving patient care. The use of AI in this field promises to automate and improve the extraction of relevant information from clinical data, enabling healthcare professionals to make more informed and personalized decisions.