October is Breast Cancer Awareness Month, a time to acknowledge the people living with breast cancer and the clinicians and researchers who work alongside them. Screening programmes exist for it, yet in some people the disease has spread by the time it is recognised.
Looking ahead, World NET Cancer Day on 10 November is a reminder that people with gastroenteropancreatic neuroendocrine tumours (GEP-NETs) typically wait nearly five years for a diagnosis. Our own work has found about 1,200 people in UK primary care records who are likely undiagnosed, on average five to seven years younger than those already diagnosed. Each of them could reach treatment sooner.
Welcome to the first Volv Global newsletter. Each edition is intended to provide you with new research and evidence you can draw on and project experiences you may put to use in your own work. Every issue will also give you a clear view of how our machine learning methodologies help pharmaceutical teams and clinicians find patients with difficult-to-diagnose diseases earlier, and predict how a diagnosed patient’s disease is likely to progress.
While we are active across diseases, this initial edition's throughline is oncology, from the timing of treatment decisions to the consequences of late diagnosis in breast cancer, with a neuroendocrine tumour case study at its heart. Further editions will follow only every two to three months.
Christopher Rudolf Chief Executive Officer, Volv Global
Featured Insight
Solid tumour care: when the decision comes before the information
In most solid tumours the patient has already been found, yet decisions are often taken before the information that would inform them is in hand. Léon van Wouwe, Clinical Innovation Director at Volv Global, shows what prediction can already do, from informing chemotherapy decisions after colon cancer surgery to estimating prognosis and prompting earlier conversations about goals. His conclusion is that prediction helps most when tied to a specific decision, and that reducing treatment may prove the larger prize.
From the archive: Beyond Initial Treatment: Navigating Recurrence Risk and Clinical Decisions in Cancer Survivors
New therapies are extending survival in some forms of breast cancer, yet their full benefit depends on reaching patients earlier. Could the early signals already held in routine health records help close that gap?
Finding GEP-NET Patients Earlier: AI-Driven Patient Detection in UK Primary Care
GEP-NETs are difficult to diagnose, with symptoms that resemble far more common conditions. Working with a leading global pharmaceutical company, Volv Global applied its machine learning methodology to about 24 million de-identified UK primary care records and found a likely-undiagnosed population equal to roughly one extra patient for every three already diagnosed. The case study sets out the method, the results and what they mean for finding these patients earlier.
Webinar recording: How AI-driven insights from RWD accelerate the diagnosis and treatment of undiagnosed patients: a case study
The same methodology flagged more than 16,000 patients who may have undiagnosed alpha-1 antitrypsin deficiency. We and our partners Takeda, Cleveland Clinic and Komodo Health explain how.
Breast cancer is well screened, yet in the United States the incidence of stage IV disease at diagnosis rose by 1.2% a year between 2010 and 2021. In England, about three in ten women invited for screening do not attend. Our disease page sets out the numbers, the symptoms and the challenges at each stage from development to launch.
Many decisions about developing a treatment rest on a patient count, and in difficult-to-diagnose diseases the diagnosed count understates the real one. inTrigue finds the patients who have a disease but no accurate diagnostic code. In hypertrophic cardiomyopathy it flagged between 2.5 and 4.6 times more likely patients than the diagnosed count. In acute hepatic porphyria, starting from a database with no confirmed diagnoses, a blinded clinician judged 90% of the top candidates plausible.
Podcast: Finding The Patients We’re Missing – Rare Rebels™ Podcast
Many undiagnosed patients are already in care. On the Rare Rebels podcast, Léon van Wouwe, our Clinical Innovation Director, explains how machine learning helps clinicians recognise them.
Vahid Esmaeili, Data Science and Digital Health Director, will co-present our poster: "Early Identification of ALS Using Advanced Machine Learning and Real-World Data" (virtually) on 29 October, at 1.30 pm ET.
Our CEO and founder Christopher Rudolf took part in the panel discussion: "From Multiomics to Clinical Intelligence: AI, Biomarkers and Precision Health" with Vaibhava Lakshmi Ravideshik (GRAIL), Fernando Leon (Kasvu Labs) and Fredrik Debong (1921.vc | CCO of OneTwenty.ai | Syntactiq.ai).
