# Digital batch records and AI process control to halve cell therapy batch failures

Source: https://onco.cc/ideas/idea-reg-ai-process-control-cell-manufacturing/  
OnCo record `idea-reg-ai-process-control-cell-manufacturing` (Idea). Data CC BY-NC 4.0, attribute "Data from OnCo (onco.cc)"; commercial use needs a licence.

## TL;DR

Autologous cell therapy batches fail more often than any other medicine because each patient's starting cells behave differently and the process runs without feedback. Inline sensors for metabolites, cell counts and cytokines, feeding models that adjust feeding and harvest timing in real time, could rescue batches that would otherwise be discarded.

## Summary

Autologous manufacturing has out-of-specification and failure rates far above those of conventional biologics, driven by variable starting material and largely open-loop processes. Inline sensors (metabolites, cell counts, imaging, cytokines), electronic batch records and machine-learning models predicting final yield and potency from early process data allow adaptive feeding, harvest timing and early re-manufacture decisions. The proposal is a shared, anonymised dataset of manufacturing runs across manufacturers and an open model benchmark, with regulatory guidance on adaptive control within a validated design space.

## Fields

- Kind: Idea
- Last checked: 2026-09-08
- Hypothesis: Adaptive control informed by shared run data reduces manufacturing failure rates by at least half and the coefficient of variation in transduced cell dose by a third, without any change in clinical safety.
- Rationale: Process analytical technology and design-space approaches transformed small-molecule and biologic manufacturing; cell therapy is the modality with the most variability and therefore the most to gain, and every run already generates the data.
- Proposed test: Pool de-identified process data from at least three manufacturers, build predictive models of failure, then run a prospective study using model-guided interventions on 200 batches versus standard operation.
- Maturity: preclinical-evidence
- Actor: engineering

## Sources

- Bottleneck evidence (Manufacturing cost and time for living and radioactive medicines): Hernandez, Prasad & Gellad, Total costs of chimeric antigen receptor T-cell immunotherapy (JAMA Oncology 2018): https://doi.org/10.1001/jamaoncol.2018.0977

## Connected records

- technologies: [CAR-T cell therapy](https://onco.cc/technologies/car-t/), [TIL therapy](https://onco.cc/technologies/til-therapy/)
- bottlenecks: [AI that is built but not validated or deployed](https://onco.cc/bottlenecks/b-ai-validation/), [Manufacturing cost and time for living and radioactive medicines](https://onco.cc/bottlenecks/b-manufacturing-cell-therapy/)
- key papers: [Total Costs of Chimeric Antigen Receptor T-Cell Immunotherapy](https://onco.cc/key-papers/paper-hernandez-jama-oncol/)

---
JSON: https://onco.cc/api/v1/entities/idea-reg-ai-process-control-cell-manufacturing.json