Avatar

Nina Deliu

Continuous Learner, Piecewise Lecturer

Statistics and Data Science

Welcome!

I am a Tenure-track Assistant Professor (RTT) in Statistics at the MEMOTEF Dept at Sapienza University of Rome, with side gigs as Visiting Faculty Researcher at Google and at MRC - Biostatistics Unit, University of Cambridge, where I was a Postdoc in 2020–2021. (Basically, I like keeping my academic calendar delightfully overbooked).

I work with collaborators all over the world — from University of Cambridge, National University of Singapore, and University of Toronto, to Italian institutions like ISTAT and NADO Italia. In previous lives, I’ve collaborated with cancer research foundations and intergovernmental associations like FAO. Along the way, I’ve picked up a few extra hats: editorial board member of YoungStatS, the blog of Young Statisticians Europe (YSE), and associate editor of Trials.

I get unreasonably excited about things like adaptive experimental designs, reinforcement learning, conformal prediction, and detecting the unexpected. I’m inspired by the messy real-world problems — and by the belief that statistical progress should be not simply good, but also good for something.

When I’m not wrangling numbers and formulas, I’m probably traveling, enjoying Italian coffee, golfing and doing yoga, or gently breaking the news that multi-armed bandits are not a pirate crew.

Education

  • PhD in Statistical Sciences, 2021

    Sapienza University of Rome

  • MSc in Statistics and Decisions, 2017

    Sapienza University of Rome

  • MSc in Mathématiques, Informatique, Décision and Organisation, 2016

    Universitè Paris Dauphine

Interests

  • STATS – uncertainty quantification, Bayes-frequentist interplay
  • ML – reinforcement learning, multi-armed bandits, conformal prediction
  • APP – biostats, design of experiments, official stats, digital education

Highlights

Our project Adaptive Experimentation Accelerator has won the 2023 XPRIZE $1M Digital Learning Challenge! We use AI and ML to improve and personalize education using adaptive experimentation.

Meet the Project & the Team

YoungStatS interview: Susan Athey

As part of the YoungStatS editorial board, I have talked with Professor Susan Athey from Stanford University. A great short read on the future of statistics, econom(etr)ics and data science!

Read the inteview here

Previous Next

Latest News

  • 🌍 (Upcoming: Jul '27) My ISR review Reinforcement Learning in Modern Biostatistics: Constructing Optimal Adaptive Interventions (with JJ Williams and B Chakraborty) turned out to be one of the journal’s most-read papers, so I got invited to the Special Invited Paper Session at the 2027 World Statistical Congress in Lusaka (Zambia). Turns out people do read long reviews.
  • 📝 (Nov '26) Out in JMIR: A Personalized Exercise Assistant using Reinforcement Learning (PEARL): Results from a Large-Scale Mobile Health Randomized-Controlled Trial, from my time as Visiting Faculty Researcher at Google: a reinforcement-learning assistant nudging people to move more. (Written, ironically, sitting down.)
  • 📝 (Oct '26) Out in JRSS-A: Scalable Generalised Accuracy Estimation for Multisource Register-based Official Statistics with the Italian National Institute of Statistics (ISTAT) team. A journey from reject and resubmit to accepted, with our most skeptical reviewer, in two quotes:
    “I don’t really like this article.”
    — Reviewer, first round
    “I was rather critical of the first version of this manuscript. […] The authors have made a genuine pedagogical effort, which is greatly appreciated.”
    — Same reviewer, final round
  • 👶 (Sep '26) The best (and smallest) news of the year: my baby girl Mia was born on 20 September, the same day as Mia Martini 🎶. What a coincidence: we decided her name before conception! I’m on maternity leave, back in late Feb ‘27, currently running an intensive longitudinal study on sleep deprivation.
  • 🎤 (Sep '26) Organised the invited session Advances in Conformal Prediction at SMPS 2026 in Lecce, then missed it: with Mia almost due, the session had to run without me. Proof that all sessions come with uncertainty.
  • 📝 (Aug '26) Out in Philosophical Transactions of the Royal Society A: The interplay between Bayesian inference and conformal prediction (with B Liseo). Bayes meets frequentism and spoiler: they get along better than you’d think.
  • 🎤 (Jun '26) Home game, big game: as part of the Local Organising Committee, I helped bring the SIS-FENStatS 2026 Joint Meeting to Sapienza, Rome. Huge effort, huge fun, little sleep (excellent training for September).
  • 🎓 (Jun '26) Now officially qualified (abilitata) as Associate Professor in Statistics (Italian National Scientific Qualification, ASN 13/D1). Waiting for tenure now!! ⏳
  • 📝 (Jun '26) New preprint with my PhD student Simone Cuonzo, on prediction intervals that care which tail you’re in: Conformal Prediction Intervals with Tail-Specific Guarantees.
  • 🎤 (May '26) Statistics with a view: I will be talking about Bayesian conformal prediction at the Approximately Bayes Workshop on the Isle of Skye (UK).
  • 📝 (May '26) Out in Science Translational Medicine: Sotatercept reduces bone morphogenetic protein signaling in patients with pulmonary arterial hypertension. The nicest reviewer comment we have ever received:
    “I don’t think I’ve ever enjoyed reviewing a paper as much as this – in part because I learned so much reading the first version and in part because the author’s response so creatively utilized input from both reviewers. I learned even more the second time.”
    — Reviewer, after revision
  • 🎤 (Mar '26) Team effort at SDS2026 in Caserta: my invited talk on Bayesian conformal prediction for small-area estimation, plus four talks of my lovely PhD students and our AUEB visitor Eva Vyltanioti.
  • 💼 (Feb '26) I became a member of the board (Giunta) of the Faculty of Economics, as well as our MEMOTEF Department, Sapienza University of Rome.
  • 🎤 (Jan '26) Conformal meets Bayes, Trieste edition: invited seminars (two joint slots with B Liseo because two is megli che uan) at DEAMS, University of Trieste.
  • 📝 (Dec '25) New preprint with my PhD student Simone Cuonzo on conformal bandits (no casinos involved!): Conformal bandits: bringing statistical validity and reward efficiency under weak arm separability.

Latest Publications

Quickly discover relevant content by filtering publications.
Nina Deliu, Piero Demetrio Falorsi, Stefano Falorsi, Diego Chianella, Giorgio Alleva (2026). Scalable Generalised Accuracy Estimation for Multisource Register-based Official Statistics. Journal of the Royal Statistical Society: Series A. DOI: 10.1093/jrsssa/qnag107

Education DOI arXiv

Nina Deliu, Brunero Liseo (2026). The interplay between Bayesian inference and conformal prediction. Philosophical Transactions of the Royal Society A, 384(2327), 20250073. DOI: 10.1098/rsta.2025.0073

Uncertainty DOI arXiv

Rajenki Das, Nina Deliu, Mark Toshner, S Sofia Villar (2026). Implementing Response-Adaptive Randomisation in Stratified Rare-disease Trials: Design Challenges and Practical Solutions. Statistical Methods in Medical Research (Special Issue on Response-Adaptive Randomization in Clinical Trials). DOI: 10.1177/09622802251380625

Adaptive inference DOI arXiv

Nina Deliu, Brunero Liseo (2026). Alternative Approaches for Estimating Highest-Density Regions. International Statistical Review, 94(1), pp. 97–120. DOI: 10.1111/insr.12592

Uncertainty DOI arXiv

See all publications

Research Projects

What I work on, who I work with, and where to start reading. Click a project to read more.

🤖 ML: Statistical reinforcement learning and multi-armed bandits

Learning the best decisions while collecting data, with applications in biostatistics, education and finance.

📏 STATS: Uncertainty quantification, from Bayesian inference to conformal prediction

How sure are we? Bayesian and conformal tools for honest uncertainty, with applications in finance and official statistics.

💊 STATS: Inference in adaptive designs

Valid tests and estimators for data collected adaptively, with applications in biostatistics and education.

🎓 APP: Advanced statistics in education

Adaptive experiments in real courses, accuracy of education statistics from registers, and models for student careers.

🧪 APP: Advanced Bayesian statistics for doping detection

Bayesian copula models that read athletes’ biological passports jointly over time, funded by the World Anti-Doping Agency.

Interested in a PhD or a research visit on one of these topics? Get in touch at nina.deliu [at] uniroma1 [dot] it with a short CV and the project that interests you.

👶 Heads-up: I am on maternity leave until late February 2027, so replies may be slower than usual.