<b>Abstract ID 90419</b> <b>Poster Board 425</b> <b>Introduction:</b> The advancement of information technologies and immediate response systems now enables the efficient collection of data on student well-being and academic performance. Such data can be instrumental in crafting personalized academic and emotional support programs. Despite this potential, these variables are often not routinely captured or utilized to enhance students' learning experiences. Our study employs artificial intelligence, specifically machine learning, to examine trends in well-being and study habits, and to explore their correlation with academic performance in a medical pharmacology course. <b>Methods:</b> In a medical pharmacology course, students completed a weekly questionnaire for 22 weeks, including mental health (WHO 5-index) and productivity queries (emotional energy quadrants), alongside time management and study methods. With IRB approval and informed consent, an anonymized dataset was used. We developed a logistic regression model with L1 regularization to predict academic performance and calculated predictive power scores (PPS) to assess the contribution of individual variables to overall well-being. The F1 score, recall, and precision metrics were used to evaluate model performance. <b>Results:</b> After conducting a pilot survey over two academic semesters, we compiled a comprehensive dataset comprising complete demographic and academic data from 74 students. The average age of participants was 20 years (range: 17-29), with 55.4% (41 students) being female. Notably, variables related to study methods and well-being evolved over the semester. After the fifth week, a trend toward mild deterioration was observed in the number of days with adequate sleep, the well-being index, and the hours spent on joyous activities. Study hours outside of lectures fluctuated, increasing significantly during midterm exam weeks. A comparison between students in high-performance and burnout emotional energy quadrants revealed that those in high-performance, on average, reported one additional day per week of sleeping at least 7 hours and engaging in 30 minutes of physical activity. Our machine learning models (logistic regression) indicated a modest predictive capability for final academic performance (grade) using behavioral or emotional variables. The most effective model attained F1-scores of 0.72 (training), 0.43 (validation), and 0.80 on the test set, with both recall and precision at 80%. Additionally, the PPS analysis identified the weekly productivity quadrant, engagement in joyful activities, and the frequency of sleeping more than 7 hours per night as the key variables impacting student mental health <b>Conclusion:</b> Our study underscores the significant potential of integrating information systems and artificial intelligence tools in monitoring students in biomedical courses. Additionally, it illuminates the diverse factors influencing academic performance, productivity, and well-being in a medical pharmacology course over an academic semester. This research represents an initial step towards utilizing analytics tools to tailor academic support plans. However, a larger dataset is essential to enhance the precision of our models. Consequently, further validation with more extensive student cohorts is recommended. The authors declare no conflicts of interest <b>Acknowledgements:</b> This study was supported by a FAPA grant from the Universidad de los Andes, awarded to Ricardo A. Peña-Silva.