Cognostix Mobile

Project introduction

Underdiagnosis of cognitive impairment is a major healthcare problem. During the pandemic, patients can’t visit doctors in person since the diagnosis is mainly done in large hospitals. These diseases range from mild cognitive impairment to more severe cases of dementia. Recently, there is also evidence of neurological problems for patients with Covid19. Current assessment for cognitive impairment is done on-site and mostly requires neuropsychological experts, which are in short supply, to get an accurate diagnosis. This project aims to build a cognitive assessment app that integrates user’s information and their background for a back-end machine learning model to predict their cognitive status.

Issue/Inspiration

Accurate diagnosis of cognitive decline is time-consuming, expensive, and not easily accessible. This makes it harder for individuals to detect it early and difficult for clinical trials to enroll qualified participants efficiently.

What it does

The app integrates users’ mobile cognitive performances, EHR data, and survey question answers, upload to secure cloud storage where machine learning models are used to accurately predict their cognitive status. It also predicts detailed performances in different cognitive domains that will be comparable to current standards and hence interpretable by medical practitioners.

How did your project evolve with the support of the COVID-19 hackathon fund by Google Cloud?

This hackathon helped us achieve most of the front-end test implementations, EHR integration, survey question formulations, and connection to secure data storage setup.

How you built it

We employed multiple open-source toolkits including CardinalKit for establishing EHR integration to secure Google Cloud storage, ResearchKit as templates to build cognitive tests.

Challenges you ran into

Communicating between people from different fields and backgrounds to build the tests and the app as everyone envisioned it.

Accomplishments you are proud of

Having most of the cognitive tests and basic user interface and flow done.

What you learned

We learned to work with people from diverse backgrounds, who bring in different expertise, in order to accomplish this app.

What’s next for your project

The project drew interest from clinicians, researchers and pharmaceutical companies. We plan to develop the app further in collaboration with our professors and their partners to eventually use in clinical trials and to aid doctors in primary care settings. To do this, we need to complete all of the intended cognitive tests, brush up the user interface, and use the app with Stanford participants who will be coming for on-site cognitive assessments in order to generate data and ground truths for machine learning model training. We’d like to integrate more data to improve the accuracy of the app using wearable (Fitbit), electronic medical records and genetics to assess and monitor risk based on gene to phenotype as a platform.

What Google Cloud products did you use to build your project?

Google Cloud Storage and Compute Engine