Apple Health: ChatGPT struggles with health data
OpenAI has launched a new feature called ChatGPT Health, which is also intended to evaluate data from Apple Health.
However, a practical test by a US journalist shows that automated health analyses should be treated with caution for now. After ChatGPT gained access to years of Apple Watch data, the system arrived at some drastic (and inaccurate) assessments of heart health. These were apparently based on a problematic weighting of individual measurements.
Incorrect evaluations of multiple parameters
The VO₂max values played a particularly important role in the evaluation. However, Apple itself points out that these are only rough estimates, primarily suitable for observing trends. Furthermore, changes in resting heart rate, which can be explained by new sensors and measurement methods when switching devices, were not correctly categorized.
The lack of consistency in the evaluations was also striking. ChatGPT produced highly variable scores for identical questions. Furthermore, basic information such as age, gender, and current health status was not consistently taken into account, even though it was available.
AI analytics for health data are not yet mature.
The test clearly demonstrates that AI-supported health analyses cannot currently replace a reliable medical assessment. Especially with sensitive topics like heart health, medical professional evaluation remains indispensable.
0 Response to "Apple Health: ChatGPT struggles with health data"
Post a Comment