Case Study
Cloud based AI Pre-Clinical trials system
Aegasis Labs partnered up with Biotech startup, Novai, to build the a platform for them. Novai was commercialising DARC Technology, an exploratory retinal biomarker for use in Age-related Macular Degeneration (AMD) & glaucoma clinical studies. DARC combines an innovative patented biologic with a state-of-the-art AI algorithm and uses standard imaging equipment to identify cellular level disease activity.
- Reliable system developed with 100% uptime and auto scaling capability
- Automated cloud-based analysis using AI. The scalable nature of the cloud-based system allowed the analysis of multiple trials to be completed concurrently.
- Automated cloud-based reporting, as live reports will be based on the collected data and are be available in real-time as the trial progresses.
Opportunity
Novai is a British biotechnology company developing and commercialising DARC technology. Novai’s immediate goal was to establish DARC (Detection of Apoptosing Retinal Cells) as the market leading biomarker in the clinical development of treatments for Glaucoma and Age-related Macular Degeneration (AMD). They won an Innovate UK grant to complete the development of this project and ship it to their first client within a year.
There was an unmet global need in the ophthalmic space, particularly in glaucoma and AMD, for accelerating and de-risking drug development. Current diagnostics in both diseases are reliant on observing structural or visual function changes, both of which take years to manifest. Using existing endpoints, clinical trials typically last between 2 – 5 years, with economic implications relating to time and cost.
COVID-19 stalled clinical trials; however, these still had to be done, hence the requirement for a remote cloud-based solution was more urgent than ever. There was more client budget available and the demand had increased as companies are looking to get back to the timelines delayed due to the disruption caused by COVID-19.
Our Solution
Aegasis Labs designed and developed the Darc Stratos Platform for Novai. The Darc Stratos platform consisted of a new proprietary biomarker and combined Artificial Intelligence (AI). The platform was built for pharmaceutical companies to enable them to de-risk and accelerate their clinical development programs, providing significant cost and time savings.
We worked with Novai to develop a Cloud based Clinical Data Trials System (CDTS) integrated with AI that allowed:
- Remote access and administration, enabling centralised data input into the clinical trials system.
- Direct logging of the participants with data directly entered into the system from global locations, thereby reducing input errors and data transfer issues.
- Automated cloud-based analysis using AI. The scalable nature of the cloud-based system allowed the analysis of multiple trials to be completed concurrently.
- Automated cloud-based reporting, as live reports were based on the collected data and are be available in real-time as the trial progresses.
Technical Bit
AI Models Training
Just like any other machine learning based system, the first step is to gather data, then label and train the AI models. Our partner had the dataset prepared and this was used to train the AI models.
Deep learning based computer vision technology was used for training of the AI models. With the usage of deep learning, the goal is to minimise human intervention and achieve human like accuracy. An ensemble of Convolutional Neural Networks were used for this task. CNNs look for patterns in the images. The first layers of the CNNs, regular features like lines/curves and edges can be detected. As we go deeper into a CNN network, more complex features are extracted which are vital for detection and classification tasks.
Using MlOps architecture to deploy AI models
The AI models were deployed using an MLOps architecture. MLOps is a set of practices that aims to deploy and maintain machine learning models in production reliably and efficiently. The key components of an MLOps solution were
- Version control to store, track and version changes for the AI code and training datasets
- Network layer to implement the network resources that ensure the MLOps solution is secured.
- ML Based workload to execute machine learning predictions
- Infrastructure as code (IaC) to automate the provisioning and configuration of cloud-based ML workloads and other IT infrastructure resources
- An ML (training/retraining) pipeline to automate the steps required to train/retrain and deploy ML models.
- A model monitoring solution to monitor production models’ performance to protect against both model and data drift. You can also use the performance metrics as feedback to help improve the models’ future development and training.
.
Cloud web app
A service based architecture was used for frontend, backend and other micro-services. Listed below are the technologies that were used for UI, UX and backend development.
- Python Tensorflow
- Python for REST API
- Angular for UI of the platform
- Docker
- Kubernetes
- Various AWS Services (Sagemaker, Elastic Container Service, Lambdas, Step Functions, AWS Batch and more…)
Results
Aegasis Labs designed, built and delivered the Darc Stratos platform with the following success metrics
- Reliable system developed with 100% uptime and auto scaling capability
- AI system with > 90% accuracy in detection early stage Glaucoma
- Ability for the platform to encrypt, store and handle patient sensitive data in different geographies
- Ability for the platform to scale using a serverless architecture
- An intuitive UI/UX that allowed multiple tiered users to perform various actions on the platform
- Automated cloud-based analysis using AI. The scalable nature of the cloud-based system allowed the analysis of multiple trials to be completed concurrently.
- Automated cloud-based reporting, as live reports will be based on the collected data and are be available in real-time as the trial progresses.
“Equipped with reputable development expertise, Aegasis Labs elevated our operations and met our success metrics by building a cloud based AI pre clinical trial system. Proactive and competent, the team independently delivered spot-on solutions based on their extensive skillset.”
John Maddison – CTO Novai
