AI-driven solutions for advanced digitalisation of ATMP quality control
| Reference number | |
| Coordinator | Högskolan i Skövde - Högskolan i Skövde Inst f biovetenskap |
| Funding from Vinnova | SEK 2 000 000 |
| Project duration | November 2024 - November 2025 |
| Status | Ongoing |
| Venture | Advanced digitalization - Enabling technologies |
| Call | Advanced and innovative digitalization 2024 - one-year projects |
Purpose and goal
The project aims to expand an AI model initially trained on complex single-cell RNA-seq data to predict cell quality of ATMPs. The new project will adapt this AI-model to work with bulk RNA data, making it suitable for broader use in standard labs without the need for advanced equipment for single cell data generation. The goal of this project is to develop a cost-effective, scalable, and digitalized QC platform for ATMP manufacturing, using human embryonic stem cells as a model system.
Expected effects and result
This project aims to significantly enhance the Swedish ATMP industry, as well as international ATMP companies, through advanced digitalisation of the QC process. By implementing efficient AI-based QC methods, the project will enable early identification of failing cell batches in ATMP manufacturing, preventing financial losses due to expensive standard QC methods and failed clinical trials.
Planned approach and implementation
The project is structured into five sequential parts: 1. Conversion of single-cell data into pseudo-bulk. 2. Development of a pseudo-bulk AI model for cell quality assessment. 3. Interpretation of transcriptomic signatures for a reduced pseudo-bulk model. 4. Transfer learning to transition from pseudo-bulk to small data. 5. AI model validation. 6. Commercialisation and dissemination of the AI solution.