Lancet Digital Health最新文献

Background

Studies on the effect of computer-aided detection (CAD) in a daily clinical screening and surveillance colonoscopy population practice are scarce. The aim of this study was to evaluate a novel CAD system in a screening and surveillance colonoscopy population.

Methods

This multicentre, randomised, controlled trial was done in ten hospitals in Europe, the USA, and Israel by 31 endoscopists. Patients referred for non-immunochemical faecal occult blood test (iFOBT) screening or surveillance colonoscopy were included. Patients were randomomly assigned to CAD-assisted colonoscopy or conventional colonoscopy; a subset was further randomly assigned to undergo tandem colonoscopy: CAD followed by conventional colonoscopy or conventional colonoscopy followed by CAD. Primary objectives included adenoma per colonoscopy (APC) and adenoma per extraction (APE). Secondary objectives included adenoma miss rate (AMR) in the tandem colonoscopies. The study was registered at ClinicalTrials.gov, NCT04640792.

Findings

A total of 916 patients were included in the modified intention-to-treat analysis: 449 in the CAD group and 467 in the conventional colonoscopy group. APC was higher with CAD compared with conventional colonoscopy (0·70 vs 0·51, p=0·015; 314 adenomas per 449 colonoscopies vs 238 adenomas per 467 colonoscopies; poisson effect ratio 1·372 [95% CI 1·068–1·769]), while showing non-inferiority of APE compared with conventional colonoscopy (0·59 vs 0·66; p<0·001 for non-inferiority; 314 of 536 extractions vs 238 of 360 extractions). AMR in the 127 (61 with CAD first, 66 with conventional colonoscopy first) patients completing tandem colonoscopy was 19% (11 of 59 detected during the second pass) in the CAD first group and 36% (16 of 45 detected during the second pass) in the conventional colonoscopy first group (p=0·024).

Interpretation

CAD increased adenoma detection in non-iFOBT screening and surveillance colonoscopies and reduced adenoma miss rates compared with conventional colonoscopy, without an increase in the resection of non-adenomatous lesions.

Funding

Magentiq Eye.

Advances in machine learning for health care have brought concerns about bias from the research community; specifically, the introduction, perpetuation, or exacerbation of care disparities. Reinforcing these concerns is the finding that medical images often reveal signals about sensitive attributes in ways that are hard to pinpoint by both algorithms and people. This finding raises a question about how to best design general purpose pretrained embeddings (GPPEs, defined as embeddings meant to support a broad array of use cases) for building downstream models that are free from particular types of bias. The downstream model should be carefully evaluated for bias, and audited and improved as appropriate. However, in our view, well intentioned attempts to prevent the upstream components—GPPEs—from learning sensitive attributes can have unintended consequences on the downstream models. Despite producing a veneer of technical neutrality, the resultant end-to-end system might still be biased or poorly performing. We present reasons, by building on previously published data, to support the reasoning that GPPEs should ideally contain as much information as the original data contain, and highlight the perils of trying to remove sensitive attributes from a GPPE. We also emphasise that downstream prediction models trained for specific tasks and settings, whether developed using GPPEs or not, should be carefully designed and evaluated to avoid bias that makes models vulnerable to issues such as distributional shift. These evaluations should be done by a diverse team, including social scientists, on a diverse cohort representing the full breadth of the patient population for which the final model is intended.

Background

Multicentre training could reduce biases in medical artificial intelligence (AI); however, ethical, legal, and technical considerations can constrain the ability of hospitals to share data. Federated learning enables institutions to participate in algorithm development while retaining custody of their data but uptake in hospitals has been limited, possibly as deployment requires specialist software and technical expertise at each site. We previously developed an artificial intelligence-driven screening test for COVID-19 in emergency departments, known as CURIAL-Lab, which uses vital signs and blood tests that are routinely available within 1 h of a patient's arrival. Here we aimed to federate our COVID-19 screening test by developing an easy-to-use embedded system—which we introduce as full-stack federated learning—to train and evaluate machine learning models across four UK hospital groups without centralising patient data.

Methods

We supplied a Raspberry Pi 4 Model B preloaded with our federated learning software pipeline to four National Health Service (NHS) hospital groups in the UK: Oxford University Hospitals NHS Foundation Trust (OUH; through the locally linked research University, University of Oxford), University Hospitals Birmingham NHS Foundation Trust (UHB), Bedfordshire Hospitals NHS Foundation Trust (BH), and Portsmouth Hospitals University NHS Trust (PUH). OUH, PUH, and UHB participated in federated training, training a deep neural network and logistic regressor over 150 rounds to form and calibrate a global model to predict COVID-19 status, using clinical data from patients admitted before the pandemic (COVID-19-negative) and testing positive for COVID-19 during the first wave of the pandemic. We conducted a federated evaluation of the global model for admissions during the second wave of the pandemic at OUH, PUH, and externally at BH. For OUH and PUH, we additionally performed local fine-tuning of the global model using the sites’ individual training data, forming a site-tuned model, and evaluated the resultant model for admissions during the second wave of the pandemic. This study included data collected between Dec 1, 2018, and March 1, 2021; the exact date ranges used varied by site. The primary outcome was overall model performance, measured as the area under the receiver operating characteristic curve (AUROC). Removable micro secure digital (microSD) storage was destroyed on study completion.

Findings

Clinical data from 130 941 patients (1772 COVID-19-positive), routinely collected across three hospital groups (OUH, PUH, and UHB), were included in federated training. The evaluation step included data from 32 986 patients (3549 COVID-19-positive) attending OUH, PUH, or BH during the second wave of the pandemic. Federated training of a global deep neural network classifier improved upon performance of models trained locally in terms of AUROC by a mean