Distribution and sharing of health data

There have been various models of sharing of health data, maintaining privacy and still maintaining HIPAA regulations. Big companies have tried to bring it all under their umbrella including Google, Apple and Amazon. However, this has not led to much success. The winner’s in this competition has usually being the software companies that offer Electronic Health record solutions that help doctors manage reimbursement. These software companies store the process the data that are required for insurance companies to manage reimbursement and have been imagining themselves as capable of storing health data. However, this has not led to any solutions.
This maybe because the healthcare data may not be all text and is mostly combination of graphics, charts, test results some imaging data and complex formatted text. This means that any operation to access the data also requires a means to process the data so that the subject who is accessing his/her data now read the data and hopefully make sense of that data. Thus, this becomes a problem in not just accessing the data but also a problem to be compatible to analyze different data formats and making sure that it is readable.
One possible solution might be a distributed approach and though this solution handles the different distributors, it does not target how the patient will be able to make sense of the data. Currently, the often-used word is “blockchain” though this is only the means to manage authenticity, many companies have tried to make this the central feature of their management of health care data
We propose another distributed solution based on how software is distributed.
The model we use is based on software distribution. Software used to be distributed through very specific pieces of hardware: Disk, CD, DVD and then with some hardware keys to make sure that the software does not get copied illegally or distributed publicly.  This software model is changing. Now software is distributed through internet connections with some authentication schemes.
In open source systems, one model of software distribution is Conda package manager (https://conda.io) that manages heterogeneity of software systems. It unifies software installations in diverse systems by describing the needs of the software by specifying meta-information as well as dependencies. This is done through a script that does the installation and reads the requirements. Amazingly this complex specification list is readable by a human and can install software across systems in diverse environments.
A similar system could be used first to manage the data and then have appropriate software available to decrypt and analyze the data that can be read. The data itself can be stored anywhere – whether with the patient, hospital or a vendor that managed the gathering of the data. The software will need location to be defined but again there can be many sources for the location. A distributed system can authenticate through blockchain or other means but it provides for security by being dependent on the specific environment of the patient/hospital asking for the data. In addition, it provides for a layer that can analyze the data.
This method for distribution has been implemented for biological software and is called Bioconda (https://bioconda.github.io). Bioconda can serve up methods and tools to analyze genomic data. Thus when we get access to our personal sequence data we will have access to analyze our sequence data rather than stare at Gigabytes of A-G-T-C in almost random order.
This is probably is not the perfect solution and is probably not ready yet, but at least it provides a way for another way to manage and understand our healthcare data.