Can we screen for pancreatic cancer? Identifying a sub-population of patients at high risk of subsequent diagnosis using machine learning techniques applied to primary care data

Abstract

BackgroundPancreatic cancer (PC) represents a substantial public health burden. Pancreatic cancer patients have very low survival due to the difficulty of identifying cancers early when the tumour is localised to the site of origin and treatable. Recent progress has been made in identifying biomarkers for PC in the blood and urine, but these cannot be used for population-based screening as this would be prohibitively expensive and potentially harmful. MethodsWe conducted a case-control study using prospectively-collected electronic health records from primary care individually-linked to cancer registrations. Our cases were comprised of 1,139 patients, aged 15-99 years, diagnosed with pancreatic cancer between January 1, 2005 and June 30, 2009. Each case was age-, sex-and diagnosis time-matched to four non-pancreatic (cancer patient) controls. Disease and prescription codes for the 24 months prior to diagnosis were used to identify 57 individual symptoms. Using a machine learning approach, we trained a logistic regression model on 75% of the data to predict patients who later developed PC and tested the model's performance on the remaining 25%. ResultsWe were able to identify 41.3% of patients < = 60 years at 'high risk' of developing pancreatic cancer up to 20 months prior to diagnosis with 72.5% sensitivity, 59% specificity and, 66% AUC. 43.2% of patients >60 years were similarly identified at 17 months, with 65% sensitivity, 57% specificity and, 61% AUC. We estimate that combining our algorithm with

Pancreatic cancer risk predicted from disease trajectories using deep learning

“…We expect further increases in prediction accuracy with the availability of data beyond disease codes, such as prescriptions, laboratory values, observations in clinical notes, diagnosis and treatment records from general practitioners (Malhotra et al 2021) and abdominal imaging (computed tomography, magnetic resonance imaging), as well as inherited genetic profiles. To achieve a globally useful set of prediction rules, access to large data sets of disease histories aggregated nationally or internationally will be extremely valuable.…”

“…For risk assessment of pancreatic cancer, recently machine learning predictive models using patient records have been built using health interview survey data (Muhammad et al 2019), general practitioners’ health records controlled against patients with other cancer types (Malhotra et al 2021), real-world hospital system data (Appelbaum, Cambronero, et al 2021; X. Li et al 2020), and from an electronic health record (EHR) database provided by TriNetX, LLC.…”

Artificial intelligence in pancreatic cancer

“…In another article published the same year, Malhotra et al used logistic regression on EHRs to screen people at high risk of PC. Their method could indicate cancer risk over a decade before diagnosing PC patients 256 . Roch et al developed a natural language processing-based pancreatic cyst identification system.…”

Development of a Model for Differentiating PDAC from Benign Pancreatic Conditions: A Prospective Case-control Study

“…14 On the other hand, even if the referrals are followed in accordance to national guidelines, chances of detecting early, potentially curable disease is miniscule, since most patients in this clinical pathway are presented with metastatic disease. 15 Predictive algorithms for identifying PDAC risk groups have been developed from large primary care databases, 16,17 mainly focusing on presenting symptoms and demographic characteristics. Results from such symptom-based cancer decision support tools (CDSTs) suggest improved discriminatory ability; 17,21 however they are over-fitted for certain patient groups and require continuous refinement with inclusion of additional features.…”

“…15 Predictive algorithms for identifying PDAC risk groups have been developed from large primary care databases, 16,17 mainly focusing on presenting symptoms and demographic characteristics. Results from such symptom-based cancer decision support tools (CDSTs) suggest improved discriminatory ability; 17,21 however they are over-fitted for certain patient groups and require continuous refinement with inclusion of additional features. 16,18 Pre-existing medical conditions are rarely considered in these algorithms and so are the associated commonly performed laboratory tests, which could enhance the performance.…”