Status: Completed

This research project began in 2019 with funding from the Pancreatic Cancer Research Fund.  The aim is to improve early detection of pancreatic cancer by leveraging historical medical records from general practices across England.

Our goal is to identify patients at higher risk of developing pancreatic cancer by applying machine learning methods to medical data. We are investigating whether certain combinations of symptoms, illnesses, or health problems occur uniquely or more frequently in individuals who are later diagnosed with this devastating disease. By identifying these symptom patterns, we hope to enable earlier triage and referral for diagnostic tests, increasing the likelihood of earlier diagnosis and more effective treatment.

Current work

Building on the foundations of our pilot study, we are now undertaking an expanded and more advanced phase of the research.

In this new study, we are using an improved approach with a larger cohort of patients and more up-to-date clinical data. Our aim is to refine and enhance the predictive tool developed in the pilot phase. Once the improved risk score is developed, we will evaluate:

• How many patients would have been recommended for screening if the tool had existed in the past
• How many cancers might have been diagnosed earlier through targeted screening
• Whether the tool is cost-effective for the NHS

This study has the potential to increase early detection rates for pancreatic cancer, ultimately improving access to timely treatment and enhancing survival outcomes.

Study outputs

Conference Presentations
 Our findings have been presented at several international conferences, including:

• Cancer in Primary Care Research International (Ca-PRI) Conference 2025,
• The Early Detection of Cancer Conference 2023,
• ESMO World Congress on Gastrointestinal Cancer (WCGIC) 2020 and published in Annals of Oncology

Publications

• Our pilot study was published in PLOS ONE in 2021.

Media Coverage

The study has received broad media attention, with coverage in:

• The Times,
• Medical Xpress,
• pharmaphorum,
• News Medical Life Sciences
• ‘In Focus’ Blog Post published online by American Journal of Managed Care (AJMC).

Other resources

• Watch our short video explaining the aims and potential impact of the pilot study.

A poster on our pilot study and current work is available here.

Status: Completed

Between 2011-2024, we were part of the twice awarded Policy Research Unit in Cancer Awareness, Screening and Early Diagnosis funded by the NIHR Policy Research Programme. As one of the seven academic institutions that form the Policy Research Unit (PRU), led by Prof. Stephen Duffy at Queen Mary University of London, we led the research stream on cancer awareness and inequalities in cancer care and management.

During the first grant award (2011-2018), our research mainly focused on explaining the large proportion of cancers diagnosed through emergency presentation in England, one of the routes to diagnosis of cancer and main indicator of diagnostic delays.  We showed that GP characteristics were not predictive of a high or low proportion of lung cancer EP (Maringe et al, 2018), and that it was mostly due to patients by-passing primary care that increased the risk of an unplanned hospitalisation.  To reduce EP, our research suggests it would require identifying the key reasons for the socioeconomic inequalities in emergency hospital admissions that persist even after controlling for the usual patient and tumour factors. (Maringe et al, 2018) We also explored the roles of cancer symptom awareness and barriers to seeking medical help on cancer outcomes. Barriers clearly vary by socio-demographics and higher barriers are associated with worse cancer outcomes (Niksic et al, 2015; Niksic et al, 2016; Niksic et al, 2016).

Throughout the final round of the programme (2019-20243), we aimed to understand whether medical reasons for hospital admission in the two years prior to the cancer diagnosis are linked to the routes to cancer diagnosis. We explored the interplay of diagnostic intensity (as reflected by patient’s history in primary and secondary care) and socioeconomic status, age, and sex.  We found that colon cancer patients diagnosed through EP, and particularly the more deprived, had the highest number of hospital emergency admissions 7 months prior to diagnosis. However, the symptoms and conditions they were admitted for were commonly vague and non-specific to colon cancer (Exarchakou et al, 2024).

To highlight the importance of tackling inequalities, we estimated the life-years lost due to inequalities. We showed that inequalities in cancer outcomes lead to more life-years lost due to cancer particularly among younger and more deprived patients (Exarchakou et al, 2022).

Status: Completed

This four-year ERC Horizon 2020 funded project (2020-2024), involved 13 institutions from 9 different countries and was led by the Hospices Civils de Lyon, France. It involved the development of a pilot platform to collect, manage and analyse a large amount of data on health status and multidimensional aspects of health-related quality of life (physical, mental and social wellbeing) of cancer patients after immunotherapy. The platform aimed to provide insights for clinicians and policy makers to propose individualised monitoring strategies following immunotherapy to minimize risks for the patients and optimise their quality of life.

Our main involvement in this project is to lead the quantitative analyses required to understand the determinants of quality-of-life (QoL) following immunotherapy. Our first study used observational real-world data to assess the impact of immunotherapy as well as other systemic cancer treatments on patients’ health related QoL who have advanced stage cancer. We also studied the role of comorbidities on QoL. In another study we described the changes in QoL of patients with advanced-stage lung cancer over 18 months following immunotherapy by different cancer treatment regimens in a real-world setting.

More specifically, to identify patterns of adverse events following immunotherapy and investigate how their occurrence impacts the different components of quality of life.

Current work

As immunotherapy improves survival for cancer patients, maintaining QoL has become increasingly important. Research shows that psychosocial factors such as social support, health literacy, trust in doctors, hope, and uncertainty can significantly influence QoL during treatment. Our current study explores how these factors impact patients’ QoL undergoing immunotherapy. Findings will help guide interventions to improve well-being alongside clinical outcomes.

While immunotherapy has transformed cancer treatment and improved survival, it can also cause serious immune-related adverse events that impact patients'QoL. Understanding these side effects is essential for ensuring safer and more patient-centred care. Our project will aim to identify patterns of adverse events following cancer immunotherapy and explore how these events affect different aspects of QoL, using electronic health records (EHR) from across England. We will also investigate whether access to these new and costly treatments vary by geography or population groups, helping to highlight potential inequalities in care.

Study outputs

Malhotra A, Suazo-Zepeda E, Vinke PC, de Bock GH, Maas WJ, Hiltermann JTJN, Rachet B, Leyrat C, Quaresma M. Changes in global quality of life after treatment with immune checkpoint inhibitors in patients receiving different treatment regimens for advanced stage lung cancer in the Netherlands: a 2015-2021 cohort study. BMJ Open. 2025 Feb 20;15(2):e098062. doi: 10.1136/bmjopen-2024-098062. PMID: 39979046

Malhotra, A., Fransen, H. P., Quaresma, M., Raijmakers, N., Versluis, M. a. J., Rachet, B., Van Maaren, M. C. & Leyrat, C. (2023). Associations between treatments, comorbidities and multidimensional aspects of quality of life among patients with advanced cancer in the Netherlands—a 2017–2020 multicentre cross-sectional study. Quality of Life Research.

Vinke, P. C., Combalia, M., Bock, G. H. D., Leyrat, C., Spanjaart, A. M., Dalle, S., Silva, M. G. D., Essongue, A. F., Rabier, A., Pannard, M., Jalali, M. S., Elgammal, A., Papazoglou, M., Hacid, M.-S., Rioufol, C., Kersten, M.-J., Oijen, M. G. V., Suazo-Zepeda, E., Malhotra, A., Coquery, E., Anota, A., Preau, M., Fauvernier, M., Coz, E., Puig, S. & Maucort-Boulch, D. (2023). Monitoring multidimensional aspects of quality of life after cancer immunotherapy: protocol for the international multicentre, observational QUALITOP cohort study. BMJ Open, 13, e069090.

Status: Completed

This three-year (2022-2025) Medical Research Council project [grant number MR/W021021/1] aims to improve the robustness of the outputs in public health research.

This is achieved through increasing the adoption of consolidated, adequate causal inference methods, making the most of recent developments in the fields of statistics, machine learning and data science, such as the targeted maximum likelihood estimation (TMLE) algorithm.

TMLE is a semiparametric double-robust, efficient substitution estimator allowing for data-adaptive estimation while obtaining valid statistical inference. In addition to being double-robust, TMLE allows the inclusion of machine learning algorithms that minimise the risk of model misspecification, a problem that persists for competing estimators.

Nonetheless, TMLE rests on relatively complex statistical and mathematical concepts that need to be demystified for wider adoption. Thus, we are developing further the user-friendly Stata command, eltmle (https://github.com/migariane/eltmle), to universalise the use of advanced causal inference techniques blending data-adaptive prediction, robust estimation, and statistical inference among applied and public health researchers.

Since 2022, we provided applied researchers with:

1. Tutorials that offer practical guidance on implementing targeted machine learning estimation using intuitive, accessible language: Targeted maximum likelihood estimation for a binary treatment: A tutorial, Introduction to computational causal inference using reproducible Stata, R, and Python code: A tutorial
2. A simple yet detailed article in the Stata Journal, and online tutorials and empirical applications illustrating the use of eltmle. [in review]
3. Demonstrations of the good properties of TMLE in simulated scenarios: Performance of Cross-Validated Targeted Maximum Likelihood Estimation
4. Real-life applications of the eltmle command

Status: Ongoing

The SORT study compares radiotherapy versus surgery for treatment of people with early-stage cancers of the lung, oesophagus and bladder, and will provide recommendations to help improve patients’ outcomes.

The main objectives are to understand:

1. the effect of having radiotherapy rather than surgery on patient’s outcomes
2. the NHS costs of radiotherapy compared to curative surgery
3. the impact of the COVID-19 pandemic on inequalities in the receipt of treatments.

The study prioritises involvement of people with experience of cancer and those from communities who have the worst outcomes. We have brought together a diverse patient and public involvement panel who meet regularly with us to feed into the research and output of results.

Further, we work with charities, such as Fight Bladder Cancer and collaborators, including the Centre for Ethnic Health Research, to reach, involve, listen, and respond to the voices of those from relevant communities.

You can read more about the project here: https://www.lshtm.ac.uk/research/centres-projects-groups/sort

The first protocol paper has been published in BMJ Open (July 2025) and is available here: https://bmjopen.bmj.com/content/15/7/e103038

Status: Ongoing

For people with cancer, NHS targets for timely treatment are not being met for surgery or radiotherapy which are the two main curative treatments for solid tumours. It is unclear how we can make better use of all available capacity for surgery and radiotherapy to provide more timely treatment. This 3.5 year NIHR funded project aims to investigate how to improve timeliness of curative surgery or radiotherapy for people with cancer by using an innovative cancer learning system model. The model will combine high quality data, innovative co-design techniques, and an inclusive online community to provide large-scale capability for understanding problems, developing solutions, and evaluating them.

This project (2025-2028) is a partnership between LSHTM and the University of Cambridge. Click here for more details.

Status: Ongoing

The disease burden of childhood cancer in Africa is among the highest globally, with a mortality-to-incidence ratio of 0.51 in Africa, compared with 0.16 in very-high Human Development Index regions. Despite being treatable and curable in high-income countries, 5-year survival is only 8% in East Africa. In the absence of comprehensive, uninterrupted data, estimating the respective roles of detection, diagnosis, and treatment in poor outcomes is challenging. A national childhood surveillance programme is required to monitor the World Health Organisation’s goal of increasing 5-year survival for childhood cancer to at least 60% by 2030.

The purpose of this observational longitudinal study is to identify and characterise children with cancer aged 0-19 years across six hospitals in Tanzania, quantify the childhood cancer burden and its determinants, and identify the key drivers of poor outcomes to support clinicians in managing children with cancer and to guide public health interventions.

This project will generate the first quantification of Tanzania’s childhood cancer burden across six hospitals within the National Children’s Cancer Network (NCNN), its determinants, and the drivers of poor outcomes, using an optimised multi-hospital electronic health record system. We will achieve this by (1) using state-of-the-art epidemiological and statistical techniques to maximise the use of available data; and (2) using a structured, systematic, and collaborative approach to co-design and co-develop actionable recommendations for sustaining childhood cancer epidemiological surveillance in Tanzania.

Status: Ongoing

Bowel cancer is one of the most common cancers in England, yet a substantial proportion of patients are diagnosed at a late stage or through emergency presentations, which are associated with poorer survival and worse patient experience. Current referral pathways rely on “red-flag” symptoms and tests such as the Faecal Immunochemical Test (FIT), but these approaches miss a proportion of cases, particularly among patients presenting with vague or non-specific symptoms. This challenge is especially pronounced among younger adults, where incidence is rising, but clinical suspicion is often low.

This Cancer Research UK–funded pilot study investigates whether machine learning methods applied to linked primary care electronic health records and national cancer registry data can help identify patients at increased risk of bowel cancer earlier in the diagnostic pathway. Using a matched case–control design, the study explores combinations of symptoms, clinical conditions, and patient characteristics recorded in routine healthcare data.  We are using advanced machine learning approaches such as Random Forests and Extreme Gradient Boosting to detect patterns that may signal increased risk before diagnosis.

The aim is to generate preliminary evidence for developing clinically useful risk prediction tools that could support earlier investigation of high-risk patients. Findings from this pilot will inform a future large-scale population-based study to develop and validate risk scores that could ultimately support clinical decision-making and improve early detection of bowel cancer.

Status: Ongoing

'Inequalities in Cancer Care and Outcomes’ is an ongoing Cancer Research UK funded Programme, initiating in 2005. Its main ambition is to reduce inequalities in cancer outcomes, ensuring that everyone, regardless of sociodemographic background, benefits from improvements in cancer care. The first three quinquennia of the programme (2005--2020) covered the following:

• measured trends and patterns of inequalities,
• identified the factors behind those inequalities and examined the role of individual tumour and patient factors on the inequalities in cancer care and survival by sociodemographic categories.

The most recent quinquennium (2020-2025), led by Professor Bernard Rachet, aims to describe how the interplay between patient characteristics and healthcare system components affect inequalities in cancer outcomes. We have recently started a fifth quinquennium (2025-2030) and will embark on an innovative evolution of our research programme focussing on a detailed understanding of the lung cancer pathway in two Cancer Alliances (Northern England and Southeast London). This programme will be co-led by Professors Bernard Rachet and Ellen Nolte, in collaboration with partners in Newcastle University and the Cancer Alliances.