At CrossCures, we are building the world’s first end-to-end AI solution for objective diagnosis and monitoring of neurological disorders.
Video Processing
The patient takes a video at the clinic during assessment or at home after the treatment’s effects kick in.
Analytics Assessment
Our platform achieves automated grading of the movement disorders by learning a stepwise approach for the respective tasks. Doctor-in-the-loop modelling is done to personalise the algorithm.
Outcomes
Our algorithm automatically grades the patient on already validated grading scales, along with a grading history. This does all the heavy lifitng a clinican has to do, but without the element of subjectivity.
Doctor-in-the-loop ML models
Models trained via a real-time feedback mechanism from expert neurologists.
Multimodal clinical data assessment
Ability of our models to connect across a plethora of modalities (videos, text, voice, biosignals, etc.).
Outcome
explainability
Our models tell the clinicians exactly what they perceive and why so.
Timeline
2022
We initially met during a hackathon and realized our shared passion for addressing the issue of subjectivity in neurological examinations. This marked the beginning of our journey, during which we dedicated two months to enhancing the existing freely available pose estimation algorithms. Our efforts paid off, as our project ultimately earned us first place in the Hack For XDP 2022 hackathon.
2022 - 2023
We focused on using our expertise to build a bespoke AI algorithm tailored specifically to approach the procedural format of clinical decision making. To introduce the doctor-in-the-loop training paradigm, we connected with leading neurologists from institutes such as Mass General, UPMC and Mayo Clinic.
June 2023
As co-founders, we were invited to the Movement Disorder Laboratory at Mass General Hospital to evaluate and reiterate our algorithm using data from real patients. After two months of continuous failures, novel innovations and rapid prototyping, we successfully achieved significant improvements in our algorithm.
August 2023
The model performed exceptionally well in these preliminary tests, achieving an accuracy rate of up to 88% with respect to expert neurologists. It demonstrated remarkable ease in handling even the most severely affected patients. With this, the 7th iteration of CrossCures as we officially begin our journey.
Co-Founder
Co-Founder
Contact Us