Personalised medicine: Humanitas pioneers the use of summarised data in clinical trials

The application of artificial intelligence (AI) in medicine has an impact that is all the more profound the more pressing the clinical needs of patients are. In this context, rare diseases – conditions with an incidence of fewer than 6 cases per 100,000 people per year, yet which collectively account for 25 per cent of cancer diagnoses – constitute a genuine public health emergency. Developing innovative therapies in these areas has historically been complex due to extreme clinical and biological heterogeneity, a lack of data and the inherent limitations of traditional randomised controlled trials (RCTs).

Traditional clinical trials, regarded as the gold standard of research, do in fact have significant limitations in the most critical contexts. Recruiting a sufficient number of patients for rare or ultra-rare conditions is often impossible due to the small size of the eligible population. Furthermore, assigning subjects with serious unmet therapeutic needs to a control arm treated with ineffective therapies or a placebo raises profound ethical dilemmas. Added to this is the under-representation of the elderly population in conventional trials and the rapid pace at which the standard of care evolves – a factor that risks rendering control arms obsolete before the study is completed.

To overcome these barriers, the cutting edge of medical research lies in synthetic data: information artificially generated by AI algorithms that faithfully replicate the statistical and biological characteristics of real data, without containing any information that can be traced back to individual patients. This application of generative AI thus makes it possible to create ‘virtual control arms’ or synthetic cohorts capable of complementing or replacing placebo groups, reducing costs and drastically shortening the validation times for an innovative treatment, as well as overcoming ethical issues related to patient recruitment.

The feasibility of this approach has been historically confirmed by the decision of the European Medicines Agency (EMA), which has approved the use of synthetic and historical data to establish the control arm in the Phase II GD2-CART01 clinical trial, conducted by the Bambin Gesù Hospital in Rome to assess the efficacy of anti-GD2 CAR-T cells in paediatric patients with relapsed or progressive neuroblastoma, a rare and highly serious cancer. This ‘green light’ sets a crucial precedent for the scientific community, demonstrating that virtual arms can effectively complement research where traditional sampling is impractical or ethically problematic.

At the forefront of this transition is the work of the IRCCS Istituto Clinico Humanitas, which has been developing synthetic data models since 2021, starting with oncohaematology and subsequently expanding into neurology, gastroenterology, hepatology and rheumatology. Through the Humanitas AI Centre – the first integrated AI research centre within an Italian IRCCS – clinicians, engineers and data scientists collaborate to create robust, reproducible and fully explainable algorithms. From this transdisciplinary ecosystem, the spin-off Train was established in 2023, focusing on the development of digital twins and synthetic datasets that remain secure within the hospital’s infrastructure. Data quality is continuously monitored by the SAFE (Synthetic vAlidation FramEwork) framework, developed as part of the European Horizon 2020 projects SYNTHIA and SYNTHEMA, which assesses the data’s statistical fidelity and clinical utility through transparent verification processes.