Description
Body temperature measurement is a low-cost and non-invasive diagnostic tool that is widely used in clinical practice to diagnose acute pathological states ranging from fever to hypothyroidism. The normal human body temperature of 98.6F was established in the mid-19th century. We used a unsupervised machine learning algorithm (LIMIT) based on the Central Limit Theorem to define normal temperature by age, sex, race, time of day and BMI. We confirmed that normal temperature is significantly lower than 98.6 in all groups and posit that LIMIT can be applied in real time to provide a standard for individual patients.
Learning Objective: After participating in the session, the learner should be better able to:
Understand how unsupervised machine learning based on the Central Limit Theorem can establish norms for commonly used clinical measures.
Authors:
Frederik Heath (Presenter)
Stanford University
Myroslava Protsiv, Stanford University
Sarah Poole, Stanford University
Julie Parsonnet, Stanford University
Presentation Materials: