Project Overview
Innodem Neurosciences specializes in neurological diagnostics using AI-driven eye-tracking technology. They partnered with us to enhance their platform, focusing on model accuracy, real-time performance on iPads, and achieving compliance with stringent regulatory standards. The project’s goal was to create a reliable tool for neurological assessments, supporting clinical use and expanding accessibility for remote patients.
Goals
1. Train and Validate AI Models:
Develop computer vision models capable of accurately assessing neurological conditions through eye-tracking data, ensuring clinical reliability and consistency.
2. Optimize Real-Time Performance:
Enhance the platform’s ability to process data instantaneously on portable, low-power devices like iPads, allowing rapid diagnostic results.
3. Ensure Regulatory Compliance:
Achieve compliance with ISO/IEC 27001, PIPEDA, and HIPAA standards while securing Health Canada certification to enable clinical use.
Challenges
1. Accuracy in Diagnostics:
Meeting the rigorous clinical standards required models that could reliably produce neurological scores with high precision. This demanded extensive training and validation against clinical data to ensure the platform’s trustworthiness.
2. Real-Time Optimization:
Delivering instant diagnostic results posed a technical challenge, especially given the need to process complex eye-tracking data on portable devices with limited computational power.
3. Regulatory Compliance:
Navigating the intricate requirements of healthcare regulations and preparing for Health Canada certification required a meticulous approach to documentation, data handling, and system design.
Solutions
1. AI Model Training and Validation:
• Robust Training Process: Developed computer vision models that analyzed eye-tracking data to generate reliable neurological assessments. These models were trained on diverse datasets to capture a wide range of patient profiles and conditions.
• Clinical Benchmarking: Benchmarked the AI models against real clinical data, ensuring their performance met or exceeded industry standards for accuracy and reliability. This process involved extensive testing to identify and resolve any discrepancies.
2. Real-Time AI Optimization:
• Streamlined Data Processing: Optimized the AI system to process eye-tracking data in real-time, allowing diagnostic scores to be delivered in under 10 seconds. This was achieved through advanced algorithmic design that minimized latency without compromising accuracy.
• Low-Power Device Adaptation: Ensured seamless functionality on iPads by refining the model architecture and resource utilization, making it both efficient and scalable for portable use.
3. Compliance and Regulatory Support:
• Adherence to Standards: Ensured the platform adhered to ISO/IEC 27001 (information security), PIPEDA (data privacy), and HIPAA (health data handling). This involved detailed audits of system architecture and data processing workflows.
• Health Canada Approval: Collaborated with regulatory experts to prepare technical submissions, ensuring the platform met all requirements for clinical certification. This included comprehensive documentation of the AI system’s reliability, security, and performance.
4. Collaboration Enablement:
• Strategic Alignment: Supported Innodem’s partnership with Novartis by ensuring technical deliverables aligned with shared goals, including advancing the innovation of neurological care solutions.
• Integrated Communication: Worked closely with all stakeholders to ensure that development milestones and compliance requirements were met without disrupting the collaboration’s broader objectives.
Results
1. Regulatory Approval:
Achieved Health Canada certification for the eye-tracking platform, allowing its use in clinical settings nationwide. This milestone validated the platform’s reliability and compliance with stringent healthcare standards.
2. Improved Accessibility:
Enabled remote self-testing for multiple sclerosis patients, with diagnostic scores delivered in under 10 minutes. This feature proved particularly valuable for rural communities, where access to neurological specialists is often limited. The tool’s portability and ease of use significantly expanded its reach and impact.
3. Validated AI Models:
The AI system delivered consistent and reliable neurological scores, proving its utility in diagnosing a range of conditions. These results reinforced the platform’s credibility and laid the foundation for broader adoption in clinical environments.
4. Strategic Growth:
Supported a successful collaboration with Novartis, allowing Innodem to strengthen its position in the neurological diagnostics field. This partnership accelerated innovation and expanded the platform’s influence in advancing patient care.
"10xStudio’s AI expertise with image processing and neural networks significantly accelerated our development speed and shortened our time to market. AI is still relatively new for upper management officers like me, and our biggest problem was communication with the AI team, which differs from other departments we usually handle. We didn’t speak the same language at all, slowing us down. Giorgio did a fantastic job explaining to management officers the steps toward achieving our goals in words we understood while ensuring the team delivered promptly and effectively. I highly recommend Giorgio and his team for their ability to understand business needs and quickly translate them into working applications."
-Yves Desgagne / CTO Innodem Neurosciences