Science Behind: Cough for certain diseases has a different pitch and frequency when seen on spectrogram. This is used to detect TB.
Patient Login with DrChrono Credentials.
TB Screening and Monitoring Dashboard for Patients.
Patient diagonised by TB and is being treated by a doctor can monitor user's cure condition by prescribing them CureSense.
Step 1 of Screening Feature[Questionarie]: Patients need to answer few criteria questions.
Step 2 of Screening Feature[Collection]: Cough sound collection from patient's mobile microphone & sending to our AI-ML servers for analysis
Step 2 of Screening Feature[Analysis]: Spectrum visualization of cough sound through mobile microphone which goes to our server for analysis
Step 3 of Screening Feature[Reports]: Our servers send the negative report for non-TB cough.
Step 3 of Screening Feature[Reports]: Our servers send the positive report for TB based cough.
Step 4: Patients can report to doctor for monitoring during treatment. On approval, it is added to patients docs using DrChrono Docs APIs.
Misc Feature: Patients can find nearby pulmonologist
Misc Feature: Users can find nearby hospitals/doctors using AR.
Misc Feature: Patients can manage their profile.
Every year around 2.2M tuberculosis(TB) cases are seen in India. In 2017, WHO recorded 0.3 million deaths in India could have been averted with early detection of the diseases.
Expensive screening systems & testing clinics far away from the villages makes it hard for villagers with chronic cough to get themselves checked for TB in early stages and often end-up taking self-mediation without a test. Which often leads to death of the patient in many cases.
Even if the patient is being treated for TB, the effectiveness of the treatment is low as the patients don’t go for periodical checkups.
What it does
As everyone has a smartphone these days, we thought why not build a system on mobile phones that could screen and monitor TB. After a lot of research we realized that cough sound or voice biomarkers vary for TB cough to other pulmonary diseases and to detect the same we came up with a mobile app CureSense.
How can tuberculosis be screened using voice biomarkers, spectrum-analysis and deep learning? Read the phenomenon here: https://www.ncbi.nlm.nih.gov/pubmed/29543189
CureSense is a mobile app which uses AI-ML coupled with voice-biomarker technique to help the users with self-screening and monitoring of tuberculosis and pulmonary diseases at early stages.
While doctors can subscribe to CureSense to their TB patients to regularly monitor the changes in cough episodes and patterns and adjust the treatment and care plan accordingly for the most effective and fastest cure.
CureSense provides 99% cheaper & faster screening for dangerous disease in both rural and urban India making it affordable to all and thus earlier detection of disease results in the direct decrease in mortality rate in the country.
How I built it
Following technologies and data were used to build CureSense:
- Android, ML-AI Deep Learning, PS, Google Cloud.
- DrChrono API: Authentication API, Document read & write API.
- Cough biomarker detection using this technical references: Tuberculosis screening using voice biomarkers, spectrum-analysis and deep learning: NCBI
Challenges I ran into
Collecting the samples for training the cough detection module was a bit challenging as we had to record or collect the cough sound from the patients from government hospitals here and it took a lot of time in training the model and making it better.
Accomplishments that I'm proud of
We were able to build the product ground up and integrate it with DrChrono in very less time, we are really happy for the opportunity.
What I learned
We learnt for the first time DrChrono api and their system. DrChrono has pretty impressive features and tools which are needed by the doctors. With right collaboration and tools like CureSense, DrChrono could be one of the smartest doctor's OS in the future.
What's next for CureSense, a mobile app based tuberculosis screening tool
We are testing our algorithms in the local community in India. We want to take it to the global level where healthcare is ineffective in identifying diseases at an early stage and build a global community of medicos and patients who need support.
Building healthcare for the future: Once our remote TB detection algorithm is optimized for efficiency, probably we would be the game-changer in the health industry to eradicate TB.