Pre-disease detection using wearable device health data

We aim for an era where you won't get sick with preventive DX!

The percentage of people with abnormal findings during regular health checkups in Japan, that is, those who are graded C or higher, is 84% among those aged 33 to 66, who are the middle generation in the 100-year lifespan era.

As the disease continues to progress dozens of times, leaving the disease untreated (a state in which the disease does not develop but is moving away from a healthy state), it eventually develops into a serious disease such as cancer, heart disease, or stroke. Resulting in. We would like to solve such problems.

We are already in an era where factory machines do not break down easily due to DX.

Just imagine, by applying such technology to humans, who are irreplaceable and even more important beings, we aim to create an era where people do not easily get sick.

Automatically acquires health data using Apple Watch, the most popular sensor among humans.

We are working on research to create an unsupervised machine learning model for user-specific anomaly detection and automatically analyze whether there are any anomalies.

We are actually trying to create a model using four years of Apple Watch health data from our research institute's representative, Yamada, and although we are still at the research stage and there are still a lot of issues to be solved, we have been able to make a number of discoveries and improvements.

If you are interested in this research, please contact us using the inquiry form. I would like to exchange opinions.

In addition, preparatory work to analyze data obtained with Apple Watch requires effort, but since there is almost no reference code released to the world, this research institute will prepare for analyzing data of Apple Watch by yourself. We have released reference code for you.

This is a simple Python code for processing i-phone healthcare data (including Apple Watch data) for visualization and machine learning. For details, please refer to GitHub below.

https://github.com/minatokuwellness/applewatch