abishekmuthian/bonehealthtracker

Bone Health Tracker

Track your Bone Health, Help create better Bone Health Treatments. - https://bonehealthtracker.com.

What

Bone Health Tracker helps us to visualize our bone health easily and helps monitor the progress of our bone disease treatment.

Why

1 in 3 women over the age of 50 years and 1 in 5 men will experience osteoporotic fractures in their lifetime¹. With 1 in 6 people in the world expected to be over the age 65 by 2050², We need more investment in understanding bone health and coming up with better treatments for bone health diseases.

How

Bone Health Tracker helps us visualize and monitor our bone health through Bone Mineral Density scan reports and also provides a way for submitting the data without any personal information for bone health research.

Demo

Clicking the above image would open the demo video in YouTube.

Features

Automatic report table generation. No need to enter confusing numbers which are prone to mistakes manually.

Skeleton showing the areas of interest dynamically from the report with different colors indicating Osteoporosis, Osteopenia and Normal bones. Helps even those without prior medical training to get better clarity on their BMD scan test reports. Legend for the colors would be displayed below the skeleton image.

T-score Chart and Z-score Chart comparing historical reports to track the progress of the treatment. Helps save lot of time for the healthcare professionals and the patients alike as this activity is usually done manually.

Change in bone density from previous year and since the start of the treatment.

Architecture

BMD report is extracted by Amazon Textract and classified by the Amazon Comprehend Medical in the python middleware.
The medical data is processed by the Go parser and any personal identifiable information is ignored.
The Parser assigns the T-scores, Z-scores and BMD scores to the individual bone sites(organs) based on the offsets.
The data is stored in a client Cookie, The Cookie is updated with new reports.
A chart canvas for T-scores and Z-scores are populated using the data from the cookie.
A skeleton image showing the areas of interest with different colors according to the findings are populated using the data from the cookie.
If the user chooses to submit the medical data for research(without any personal information), The cookie is updated with additional data, the data is then converted to JSON and stored in a sqlite3 database.

Testing Setup

Requirements

Go version 1.19.2 (or) higher.
Python version 3.10.6 (or) higher.

Database

Create a sqlite3 database bht.db inside the db folder.

$ cd db 
$ touch bht.db

Middleware Server

Install the required Python packages from requirements.txt.

Install uvicorn

$ pip3 install "uvicorn[standard]"

Setup Amazon Textract and Amazon Comprehend Medical using CLI.

Run the middleware service in the bonehealthtracker folder,

 $ cd bonehealthtracker
 $ uvicorn main:app --host 0.0.0.0 --port 8000

Note:

In production, I use gunicorn with multiple workers for greater performance as found in /backup/server_scripts/gunicorn.service. For using gunicorn, It needs to be installed along with uvicorn as detailed earlier.

Configuration

The configuration file is located in /secrets/fragmenta.json. SQL schema is located in db/Create-Tables.sql.

Information on some of the configuration fields -

db: Location of the sqlite3 bht.db.
port: Port for the application, 3000 is set by default in the development configuration.
classifier_server: Middleware Server created earlier, e.g. [Middleware Server IP]:8000/classification.

Building the application

After ensuring Go is installed in the system,

    $ cd bonehealthtracker
    $ go build bonehealthtracker

Run the application

After ensuring Middleware Server is running,

$ ./bonehealthtracker

Usage

Visit the Bone Health Tracker project domain you have set in root_url with set port e.g. localhost.com:3000.
Upload the assessment section of the Bone Mineral Density(BMD) scan test report.

Note: FakeData folder contains BMD sample reports for testing.
- Assessment-1.jpg to Assessment-3.jpg are of standard format and can be uploaded in sequence to test the visualizations for Osteoporosis, Osteopenia & Normal bone density.
- BDRS-1.png is of medical transcription type with Osteopenia.
- BDRS-2.png is another medical transcription type with normal bone density.

Miscellaneous Testing Notes

Occasionally, Amazon Comprehend Medical returns incorrect offset for the BMD scores and thereby mismatching the BMD score for those bone sites. Deleting the report and Uploading it again fixes the issue.
Bone Health Tracker can currently display utmost 3-4 reports due to limitations with cookie size.
If a later report contains a new bone site, Then it could be shown as belonging to a wrong year. Usually, Consecutive reports have same bone sites for a patient.
Bone Health Tracker uses Cloudflare Turnstile for the security of the forms as an alternative for captcha. This is necessary because Bone Health Tracker could be used by elderly and the disabled where a captcha can cause friction. Turnstile is currently in Beta and so there are rare occasions when the widget doesn't get displayed; A browser refresh usually fixes it and if doesn't then we need to give a gap of 10-15 mins before attempting again.
Parser prints error when it encounters elements it couldn't parse, These are expected and doesn't affect the function of the application.

License

This program is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more details.

You should have received a copy of the GNU Affero General Public License along with this program. If not, see https://www.gnu.org/licenses/.