The Alan Turing Institute – Roche strategic partnership

Introduction

The Turing's health and medical sciences programme has long-term interests in personalised healthcare and treatment heterogeneity – the analysis of how medical treatments can affect different people in different ways. The programme is seeking to generate insights that improve the understanding of patient and disease heterogeneity and its relevance to clinical outcomes.

The goal of this partnership is to establish a world-leading collaboration in advanced analytics between Roche and the Turing, focused on enabling the transformative benefits of personalised healthcare to become a reality for patients around the world. Publication of methods and algorithms will follow the principles of open science to ensure that they are reproducible and interoperable. 

Read the news story: The Alan Turing Institute launches a strategic partnership with Roche to generate insights into disease, patient and outcome heterogeneity using advanced analytics

About Roche

Learn more about Roche

Background

This expanded strategic partnership was launched following a successful Data Study Group at the Turing in April 2019 and a 12-week research project later the same year, explored in the case study Turing and Roche: Towards tailor-made lung cancer treatment.

More information on the partnership can be found in this short video from our event hosted in July 2021:

Current projects

The Turing-Roche partnership is running for five years and will cover multiple activities with the "North Star" of developing new data science methods to investigate large, complex, clinical and healthcare datasets to better understand how and why patients respond differently to treatment, and how treatment can be improved. The partnership is currently pursuing three research themes, which you can find out more below.

Structured missingness in heterogeneous data

Missing data is a common problem that arises in many fields and can significantly complicate analysis. While there are established ways to handle data that is missing at random, it is often the case that missing readings are structured in some way. Dealing with such structured missingness is substantially more challenging yet occurs commonly in healthcare (and other) contexts. Consequently, there is a need to develop rigorous ways to understand structured missingness and develop tools to handle it appropriately. 

In November 2021, we ran a series of virtual workshops on the topic of structured missingness, with the aim of convening a community of researchers interested in developing new tools and methods, to learn from heterogenous data with structured patterns of missing readings. You can read more about the workshops in the report in the downloads section below. We then ran a competitive funding call in this area to complement the workshops and fund projects in this area. We funded two 18-month projects to Professor Ginestra Bianconi and Dr Robin Mitra and their teams which you can read more about here. These projects started in September 2022 and we will be sharing regular updates from the projects through our Slack Workspace and Knowledge Share events, as well as open publications of methods, algorithms and results. From August 2024, Research Associate Viktor Racinskij is also working within this workstream.

A blog about our partnership and this area of research was published on Genomics England's site Mind the Gap: Stories of Health Data Equity. The site publishes stories on different ideas, initiatives and perspectives to showcase the challenges of data diversity. You can read our piece here.

Furthermore, the partnership has also published a perspectives piece on structured missingness in Nature Machine Intelligence. A product of the collaborative workshops mentioned above, the publication explores challenges that address and mitigate effects of structured missingness and where current research could go from here, as well as creating tools to tackle structured missingness; gaining a more holistic understanding of it could convey important information about the data itself. You can find the publication here.

Predictive modelling

We are exploring the research theme of predictive modelling in healthcare and how this branch of advanced analytics can potentially predict and help improve patient outcomes and beyond. In October 2022, we ran a collaborative workshop with researchers to explore this theme; you can read about it in our downloads section. We also ran an associated funding call and awarded funding for an 18-month project to Dr Brieuc Lehmann and his team (Dr Karla Diaz-Ordaz, Dr Matthew Sperrin, Prof Ricardo Silva and Dr Ioanna Thoma); read more about this here.

We also have two Research Associates under this theme. Enrico Parisini is working on a project on Geometry of Deep Learning (previously worked on by Alessandro Barp) and Ariane Marandon is working on a project on Conformal Prediction.

The partnership has published an article on predictive uncertainty in clinical AI tools in Nature Medicine, discussing that, while AI tools usually aim to maximise predictive accuracy, personalised measures of uncertainty - using new techniques such as conformal prediction - are needed for clinical AI to realise its potential and improve human health. You can read it here.

Multimodal analysis

This research theme of the partnership is exploring how we can most effectively combine different modalities of data to improve clinical care. Activities for this theme will be launched in early 2024; we are currently exploring potential research projects using the The Cancer Genome Atlas (TCGA) dataset. We have recruited Fiona Young as Research Assistant to work on this theme, alongside colleagues from the Turing Research Data Team, Luis Santos and Mahwish Mohammad.

Publications

Featured publications:

• Mitra R., McGough S.F., Chakraborti T. et al. Learning from data with Structured Missingness, Nature Machine Intelligence, 25 January 2023 [open access version here]
• Sun H., Radicchi F., Kurths J. et al.  The dynamic nature of percolation on networks with triadic interactions, Nature Communications, 10 March 2023
• Cowley M.V., Pruller J., Ganassi M. et al. An in silico FSHD muscle fiber for modeling DUX4 dynamics and predicting the impact of therapy, eLife, 15 May 2023
• Das N., Saha S., Nasipuri M. et al. Deep-Fuzz: A synergistic integration of deep learning and fuzzy water flows for fine-grained nuclei segmentation in digital pathology, PLOS One, 23 June 2023
• Jackson J., Mitra R., Hagenbuch N. et al. A Complete Characterisation of Structured Missingness, arXiv, 05 July 2023
• Banerji C.R.S., Chakraborti T., Harbron C. et al. Clinical AI tools must convey predictive uncertainty for each individual patient, Nature Medicine, 11 October 2023 [open access version here]
• Baptista A., MacArthur B.D., Banerji C.R.S. Charting cellular differentiation trajectories with Ricci flow, Nature Communications, 13 March 2024
  • LinkedIn Blog about paper here
  • Springer Nature Blog about paper here
• Paul D., Saha S., Basu S. et al. Computational analysis of pathogen-host interactome for fast and low-risk in-silico drug repurposing in emerging viral threats like Mpox, Nature Scientific Reports, 12 August 2024
• Baptista A., Barp A., Chakraborti T. et al. Deep Learning as Ricci Flow, Nature Scientific Reports, 08 October 2024
  • Springer Nature blog about paper here
• Banerji C.R.S., Chakraborti T. et al. Train clinical AI to reason like a team of doctors, Nature Comments, 04 March 2025

Other publications:

• Calmon L., Schaub M.T., Bianconi G. et al. Higher-order signal processing with the Dirac operator, IEEE, 07 March 2023
• Banerji C.R.S., Heher P., Hogan J. et al. Symptom onset and cellular pathology in facioscapulohumeral muscular dystrophy is accelerated by cigarette smoking, medRXiv, 24 May 2023
• Paul S., Das N., Dutta S.B. et al. LSIP: Locality Sensitive Intensity Projection for 2D Mapping of High-Res 3D Images of Dendritic Spines, International Conference on Frontiers in Computing and Systems (COMSYS 2022 oral paper), 01 August 2023
• Guérin A., Basu S., Chakraborti T. et al. Deep Visualisation-Based Interpretable Analysis of Digital Pathology Images for Colorectal Cancer, International Conference on Frontiers in Computing and Systems (COMSYS 2022 oral paper), 01 August 2023
• Banerji C.R.S., Greco A., Joosten L.A.B. et al. The FSHD muscle–blood biomarker: a circulating transcriptomic biomarker for clinical severity in facioscapulohumeral muscular dystrophy, Brain Communications, 16 August 2023
• Engquist E.N., Greco A., Joosten L.A.B. et al. FSHD muscle shows perturbation in fibroadipogenic progenitor cells, mitochondrial function and alternative splicing independently of inflammation, Human Molecular Genetics, 19 October 2023
• Knott A., Pedreschi D., Chatila R. et al. Generative AI models should include detection mechanisms as a condition for public release, Ethics and Information Technology, 28 October 2023
• Sriwastava B., Halder AK., Basu S. et al. RUBic: rapid unsupervised biclustering, BMC Bioinformatics, 16 November 2023
• Baptista A., Sánchez-García R.J., Baudot A. et al. Zoo Guide to Network Embedding, Journal of Physics: Complexity, 20 November 2023
• Dey S., Basuchowdhuri P., Mitra D. et al. BliMSR: Blind Degradation Modelling for Generating High-Resolution Medical Images, Medical Image Understanding and Analysis (conference paper), 02 December 2023
• Nguyen T-H., Paprzycki L., Legrand A. et al. Hypoxia enhances human myoblast differentiation: involvement of HIF1α and impact of DUX4, the FSHD causal gene, Skeletal Muscle, 16 December 2023
• Saha S., Chatterjee P., Nasipuri M. et al. Computational drug repurposing for viral infectious diseases: a case study on monkeypox, Briefings in Functional Genomics, 05 January 2024
• Dey S., Chakraborti T., Basuchowdhuri P. et al. MoMSGAN: Mode Collapse based Degradation Agnostic Multi-Scale Super-Resolution of Medical Images, Association for Computing Machinery (conference paper), 31 January 2024
• Zhan S., Chakraborti T., Wang W. et al. Uncertainty quantification of cuffless blood pressure estimation based on parameterized model evidential ensemble learning, Biomedical Signal Processing and Control, 21 February 2024
• Banerjee R., Saha SK., Chakraborti T. et al. Ret2Ret: Retinal Blood Vessel Extraction via Improved Pix2Pix Image Translation, Lecture Notes in Electrical Engineering (conference paper), 06 March 2024
• Nguyen T-H., Limpens M., Bouhmidi S. et al. The DUX4–HIF1α Axis in Murine and Human Muscle Cells: A Link More Complex Than Expected, International Journal of Molecular Sciences, 15 March 2024
• Chakraborti T. How Human-Centered Are the AI Systems That Implement Social Media Platforms?, in Regis C. (ed), Human-Centered AI: A Multidisciplinary Perspective for Policy-Makers, Auditors, and Users, 22 March 2024
• Blanchard G., Durand G., Marandon A. et al. FDR control and FDP bounds for conformal link prediction, arXiv, 03 April 2024
• Baptista A., Niedostatek M., Yamamoto J. et al. Mining higher-order triadic interactions, arXiv, 23 April 2024
• Engquist E.N, Greco A., Joosten L.A.B. et al Transcriptomic gene signatures measure satellite cell activity in muscular dystrophies, iScience, 08 May 2024
• De A., Das N., Saha PK. et al. MSO-GP: 3-D segmentation of large and complex conjoined tree structures, Methods, 07 June 2024
• Asad E., Mollah AF., Basu S. et al Deep features and metaheuristics guided optimization-based method for breast cancer diagnosis, Multimedia Tools and Applications, 28 June 2024
• Qian L., Wang T., Wang J. et al. How Deep is your Guess? A Fresh Perspective on Deep Learning for Medical Time-Series Imputation, arXiv, 11 July 2024
• Knott A., Pedreschi D., Jitsuzumi T. et al. AI content detection in the emerging information ecosystem: new obligations for media and tech companies, Ethics and Information Technology, 21 September 2024
• Ansari F., Chakraborti T., Das S. Algorithmic Fairness in Lesion Classification by Mitigating Class Imbalance and Skin Tone Bias, Medical Image Computing and Computer Assisted Intervention – MICCAI 2024 (Conference Paper), 03 October 2024
• Ren Z., Li Y., Li X. et al. SkinCON: Towards Consensus for the Uncertainty of Skin Cancer Sub-typing Through Distribution Regularized Adaptive Predictive Sets (DRAPS), Medical Image Computing and Computer Assisted Intervention – MICCAI 2024 (Conference Paper), 03 October 2024
• Dey S., Kundu B., Basuchowdhuri P. et al. A Fast Domain-Inspired Unsupervised Method to Compute COVID-19 Severity Scores from Lung CT, Intl Conf on Pattern Recognition (ICPR), 04 December 2024
• Banerjee R., Mujib R., Sanyal P. et al. Self-supervised retinal disease analysis, Medical & Biological Eng & Computing, 14 December 2024
   

Opportunities and engagement 

Knowledge Share Series

Our knowledge share series aims to bring together members of both Roche and the Turing's networks and the wider scientific community to showcase partnership updates and research, hear different academic and industry perspectives on data science topics and help build new connections and collaborations. Find our monthly events and how to register here.

Newsletter

Sign up to The Alan Turing Institute-Roche strategic partnership mailing list to receive regular updates from the strategic partnership team and community. Mailings include information about Turing-Roche projects and opportunities, jobs, events and funding calls from both the Alan Turing Institute and Roche. If the link above does not work for you (i.e. blocked by Roche's firewall), then please sign up via this form.

Turing-Roche Slack Workspace

Request to join our Slack Workspace to connect with the partnership team and the wider community.

YouTube Channel

Catch up with recordings of our previous events and partnership videos at our YouTube channel.

GitHub Organisation

The partnership has a GitHub Organisation where you'll find project and community related repositories and documentation from our partnership.

Community Scholar Scheme

The partnership Community Scholar Scheme launched in September 2023. This year-long scheme supported 10 PhD students and awarded them a stipend of £3,000 to embed themselves within the partnership community, attend a relevant conference and undertake a selected, community-based project. The scheme gave the scholars experience of this unique academic-industry partnership and allowed them to develop skills that will help further and benefit their careers. The community projects run by the scholars gave the partnership community and wider interested researchers further opportunities to engage in topics relevant to the partnership and data science and health more broadly. 

You can find out more about how the scheme ran and all of the scholars outputs in the 2023/24 Yearbook.