Healthcare research supported by Digital Planning will be showcased at an international AI conference in Spain.
The paper, produced by Digital Planning’s PhD researcher Aman Wakade in partnership with University of Derby, tackles how hospitals use AI to improve patient outcomes without compromising data privacy.
It will be presented in person at the Federated Learning and Intelligent Computing Systems Conference (FLICS 2026) in Valencia in June.
Published through IEEE, the world’s leading professional body for technology and engineering, the research will reach a global audience of practitioners and academics.
The research demonstrates how hospitals can collaboratively train a powerful transplant risk prediction model that protects patient privacy, yet delivers performance comparable to conventional, fully centralised AI systems.
Secure AI research to improve hospital decision-making
Hospitals hold valuable data that could save lives but can’t be used due to sensitive, legally protected client records that are siloed across different institutions.
The problem is that sharing data centrally creates unacceptable privacy risks while working from isolated datasets limits what AI can learn to solve the problem.
This research tests a way around that. Each hospital trains an AI model on its own data only, sharing just the patterns it identifies rather than the underlying records.
A central system combines those patterns into a smarter overall model with no patient data ever leaving the institution.
Using over 1.2 million real transplant records from the Organ Procurement and Transplantation Network (OPTN) registry, the study simulates five hospital centres.
It then trains an AI model to predict 1‑year graft survival, a key outcome shaping post‑transplant recovery, follow‑up and long‑term success.
The results show that this privacy-first federated approach matches the performance of centralised systems while keeping all patient data within each hospital’s own systems.
That means hospitals can collaborate to build more accurate decision‑support tools without weakening their data protection or changing how sensitive information is stored.
The work builds directly on Digital Planning’s wider focus on federated learning, training AI models across silos without compromising privacy or data classification, as a route to secure, scalable and decentralised intelligence.
These concepts are important in data‑sensitive sectors such as healthcare, finance and defence – where traditional centralised AI can be difficult to deploy.
Chief Science Officer Ben Hutchings, said: “By advancing federated learning, we’re unlocking industries previously constrained by data sensitivity – and doing it in a way that’s secure, scalable, and practical.”
Transparent risk insights
The research includes tools that detail why the model has flagged a particular risk – including clinical factors such as donor age and kidney quality scores.
By using this model clinicians can assess post‑transplant risk earlier, triage complex cases consistently, and justify their decisions with clear, data-led reasoning.
This paper forms part of Aman’s four year PhD programme co-sponsored by Digital Planning and the University of Derby, this is another conference paper he wrote Recognizing Cardiovascular Risk Patterns using Ensemble Learning Algorithms.
