I Am A Datapoint . org
The IADP Foundation is a non-profit filed in Massachusets in 2025.
Organization By-laws Article I: Purpose:
1) To grow a dataset over time that can aid in the understanding of human behavior and mental health.
2) To develop an engaging, adaptive platform for collecting high quality self-reported phenotypic data from as many users as possible, while encouraging continued engagement (and ideally, longitudinal data).
3) The foundation will eventually try to organize volunteer-based studies to collect objective data (from wearable devices, for instance) to pair with the phenotypic data in order to conduct research.
4) The foundation will also look to partner with other organizations, or license the data, to enable biomarker or biotyping research.
More background:
This idea for this project came from working in this field as a data scientist for the last decade. I spent 3 years in industry and 5 at a non-profit, working to understand and improve the treatment of various medical conditions, often by way of biomarker discovery.
In one case, I worked with a dataset of about 800 partipants who had been profiled for anxiety, depression, and trauma with about a 30 different symptom scales, and with a dozen cutting-edge biology platforms like whole genome sequencing, RNAseq, metabolomics, fMRI, EEG, etc. And even as someone with a background in both biology and machine learning, it seemed like a mistake to be focusing on all the biology. It would be much easier, much cheaper, and much more effective, long-term, to focus on the phenotype data - the symptom scales and clinical-work-ups. With 800 people, you could already see some really interesting and potentially actionable trends in that data . But if you could collect data from 8,000 people, and get 10% percent of them to come check back in for longitudinal data, that kind of dataset could yield huge insights.
Oddly, this was a project/program at the non-profit I worked at, but it was clearly driven by a for-profit motive - measuring stuff in the blood is something you can make money doing. Asking people the perfect combination of 10 behavioral health questions isn't, really.
I look around and see hundreds of millions of dollars being spent toward the goal of developing "biomarkers" to improve treatment of human diseases - even psychiatric conditions. People in that industry love to talk about the power of AI and machine learning, but with even a basic understanding of deep learning, one realizes the sample sizes needed for that aren't the kind of sample sizes collected (in academia or industry) when it comes to the latest greatest 'omics' platform. Even large scale efforts attempting to change this trend are numbered in the thousands.
here.
1) To grow a dataset over time that can aid in the understanding of human behavior and mental health.
2) To develop an engaging, adaptive platform for collecting high quality self-reported phenotypic data from as many users as possible, while encouraging continued engagement (and ideally, longitudinal data).
3) The foundation will eventually try to organize volunteer-based studies to collect objective data (from wearable devices, for instance) to pair with the phenotypic data in order to conduct research.
4) The foundation will also look to partner with other organizations, or license the data, to enable biomarker or biotyping research.
In one case, I worked with a dataset of about 800 partipants who had been profiled for anxiety, depression, and trauma with about a 30 different symptom scales, and with a dozen cutting-edge biology platforms like whole genome sequencing, RNAseq, metabolomics, fMRI, EEG, etc. And even as someone with a background in both biology and machine learning, it seemed like a mistake to be focusing on all the biology. It would be much easier, much cheaper, and much more effective, long-term, to focus on the phenotype data - the symptom scales and clinical-work-ups. With 800 people, you could already see some really interesting and potentially actionable trends in that data . But if you could collect data from 8,000 people, and get 10% percent of them to come check back in for longitudinal data, that kind of dataset could yield huge insights.
Oddly, this was a project/program at the non-profit I worked at, but it was clearly driven by a for-profit motive - measuring stuff in the blood is something you can make money doing. Asking people the perfect combination of 10 behavioral health questions isn't, really. I look around and see hundreds of millions of dollars being spent toward the goal of developing "biomarkers" to improve treatment of human diseases - even psychiatric conditions. People in that industry love to talk about the power of AI and machine learning, but with even a basic understanding of deep learning, one realizes the sample sizes needed for that aren't the kind of sample sizes collected (in academia or industry) when it comes to the latest greatest 'omics' platform. Even large scale efforts attempting to change this trend are numbered in the thousands. here.