Breastfeeding has health benefits for infants and mothers, yet the UK has low rates with marked social inequalities. The Assets-based feeding help Before and After birth (ABA) feasibility study demonstrated the acceptability of a proactive, assets-based, woman-centred peer support intervention, inclusive of all feeding types, to mothers, peer supporters and maternity services. The ABA-feed study aims to assess the clinical and cost-effectiveness of the ABA-feed intervention compared with usual care in first-time mothers in a full trial.

A multicentre randomised controlled trial with economic evaluation to explore clinical and cost-effectiveness, and embedded process evaluation to explore differences in implementation between sites. We aim to recruit 2730 primiparous women, regardless of feeding intention. Women will be recruited at 17 sites from antenatal clinics and various remote methods including social media and invitations from midwives and health visitors. Women will be randomised at a ratio of 1.43:1 to receive either ABA-feed intervention or usual care. A train the trainer model will be used to train local Infant Feeding Coordinators to train existing peer supporters to become ‘infant feeding helpers’ in the ABA-feed intervention. Infant feeding outcomes will be collected at 3 days, and 8, 16 and 24 weeks postbirth. The primary outcome will be any breastfeeding at 8 weeks postbirth. Secondary outcomes will include breastfeeding initiation, any and exclusive breastfeeding, formula feeding practices, anxiety, social support and healthcare utilisation. All analyses will be based on the intention-to-treat principle.

The study protocol has been approved by the East of Scotland Research Ethics Committee. Trial results will be available through open-access publication in a peer-reviewed journal and presented at relevant meetings and conferences.

ISRCTN17395671.

Early identification of lung cancer on chest radiographs improves patient outcomes. Artificial intelligence (AI) tools may increase diagnostic accuracy and streamline this pathway. This study evaluated the performance of commercially available AI-based software trained to identify cancerous lung nodules on chest radiographs.

This retrospective study included primary care chest radiographs acquired in a UK centre. The software evaluated each radiograph independently and outputs were compared with two reference standards: (1) the radiologist report and (2) the diagnosis of cancer by multidisciplinary team decision. Failure analysis was performed by interrogating the software marker locations on radiographs.

5722 consecutive chest radiographs were included from 5592 patients (median age 59 years, 53.8% women, 1.6% prevalence of cancer).

Compared with radiologist reports for nodule detection, the software demonstrated sensitivity 54.5% (95% CI 44.2% to 64.4%), specificity 83.2% (82.2% to 84.1%), positive predictive value (PPV) 5.5% (4.6% to 6.6%) and negative predictive value (NPV) 99.0% (98.8% to 99.2%). Compared with cancer diagnosis, the software demonstrated sensitivity 60.9% (50.1% to 70.9%), specificity 83.3% (82.3% to 84.2%), PPV 5.6% (4.8% to 6.6%) and NPV 99.2% (99.0% to 99.4%). Normal or variant anatomy was misidentified as an abnormality in 69.9% of the 943 false positive cases.

The software demonstrated considerable underperformance in this real-world patient cohort. Failure analysis suggested a lack of generalisability in the training and testing datasets as a potential factor. The low PPV carries the risk of over-investigation and limits the translation of the software to clinical practice. Our findings highlight the importance of training and testing software in representative datasets, with broader implications for the implementation of AI tools in imaging.