Early Experiences of Integrating an Artificial Intelligence-Based Diagnostic Decision Support System into Radiology Settings: A Qualitative Study.

BACKGROUND: Artificial Intelligence (AI) based clinical decision support systems to aid diagnosis are increasingly being developed and implemented but with limited understanding of how such systems integrate with existing clinical work and organizational practices. We explored the early experiences of stakeholders using an AI-based e-learning imaging software tool Veye Lung Nodules (VLN) aiding the detection, classification, and measurement of pulmonary nodules in computed tomography scans of the chest. We performed semi-structured interviews and observations across early adopter deployment sites with clinicians, strategic decision-makers, suppliers, patients with long-term chest conditions, and academics with expertise in the use of diagnostic AI in radiology settings. We coded the data using the Technology, People, Organizations and Macro-environmental factors framework (TPOM). We conducted 39 interviews. Clinicians reported VLN to be easy to use with little disruption to the workflow. There were differences in patterns of use between experts and novice users with experts critically evaluating system recommendations and actively compensating for system limitations to achieve more reliable performance. Patients also viewed the tool positively. There were contextual variations in tool performance and use between different hospital sites and different use cases. Implementation challenges included integration with existing information systems, data protection, and perceived issues surrounding wider and sustained adoption, including procurement costs. Tool performance was variable, affected by integration into workflows and divisions of labor and knowledge, as well as technical configuration and infrastructure. These under-researched factors require attention and further research.

Early experiences of integrating an artificial intelligence-based diagnostic decision support system into radiology settings: a qualitative study.

OBJECTIVES: Artificial intelligence (AI)-based clinical decision support systems to aid diagnosis are increasingly being developed and implemented but with limited understanding of how such systems integrate with existing clinical work and organizational practices. We explored the early experiences of stakeholders using an AI-based imaging software tool Veye Lung Nodules (VLN) aiding the detection, classification, and measurement of pulmonary nodules in computed tomography scans of the chest. MATERIALS AND METHODS: We performed semistructured interviews and observations across early adopter deployment sites with clinicians, strategic decision-makers, suppliers, patients with long-term chest conditions, and academics with expertise in the use of diagnostic AI in radiology settings. We coded the data using the Technology, People, Organizations, and Macroenvironmental factors framework. RESULTS: We conducted 39 interviews. Clinicians reported VLN to be easy to use with little disruption to the workflow. There were differences in patterns of use between experts and novice users with experts critically evaluating system recommendations and actively compensating for system limitations to achieve more reliable performance. Patients also viewed the tool positively. There were contextual variations in tool performance and use between different hospital sites and different use cases. Implementation challenges included integration with existing information systems, data protection, and perceived issues surrounding wider and sustained adoption, including procurement costs. DISCUSSION: Tool performance was variable, affected by integration into workflows and divisions of labor and knowledge, as well as technical configuration and infrastructure. CONCLUSION: The socio-organizational factors affecting performance of diagnostic AI are under-researched and require attention and further research.

Feasibility of laparoscopic tumour nephrectomy in children.

AIMS: The laparoscopic approach to tumour nephrectomy in children is controversial. We therefore reviewed our institution's cases of tumour nephrectomy (laparoscopic, open, and converted) to better understand which is suitable for this approach, what factors prevent it, and whether one can excise tumours greater than the CCLG recommendation of 300 ml. METHODS: All tumour nephrectomies performed between 2002 and 2016 were identified using our surgical database. Further data were gathered from radiology and pathology databases. Those with nonrenal tumours or having a partial nephrectomy were excluded. Tumour maximum diameters, volumes, and ratios to contralateral kidneys were calculated. A Mann-Whitney U was used to compare the groups. RESULTS: Forty-three cases were included. Fifteen procedures were completed laparoscopically (35%), and a further 3 converted. The median age at surgery was 2.5 years (range 0-10) in the laparoscopic group and 2 years (range 0-15) in the open group. There was a significant difference (P < 0.05) between the laparoscopic and open groups for: median maximum diameter (10cm vs 12.25cm), median volume (155 ml vs 459 ml), maximum diameter ratio (1.22 vs 1.75), and volume ratio (3.8 vs 11.2). CONCLUSION: Tumours in the laparoscopic group were significantly smaller, but it was possible to excise tumours more than 300 ml. Difficulties in excision related to tumour size relative to the abdomen. Therefore, a ratio of tumour to contralateral kidney may be a better guide to safe excision than an overall volume cutoff. From our series, the laparoscopic approach is likely to be achievable if the volume ratio is ≤ 8.1. LEVEL OF EVIDENCE: Level 3.

Microbiological characteristics of acute osteoarticular infections in children.

This study aimed to describe the microbiological characteristics of acute septic arthritis (SA) and osteomyelitis (OM) in children. Cases of children (0-15 years) with SA/OM were identified through a retrospective search of hospital discharge codes over a six-year period. In addition, a systematic literature search and meta-analysis of studies reporting culture results of children with SA/OM was performed. In our retrospective chart review, we identified 65 cases of OM and 46 cases of SA. The most frequently cultured organisms in both conditions were Gram-positive cocci, primarily Staphylococcus aureus. On admission, most patients had a normal white blood cell count (WCC) but elevated C-reactive protein (CRP) and/or erythrocyte sedimentation rate (ESR). Bacteraemia was associated with a longer mean length of hospitalization for both infections. Considering our results and the meta-analysis, we found low rates of culture-positivity in cases of clinically confirmed infection. In SA, articular fluid was culture-positive in 42.49% [95% confidence interval (CI) 28.39-57.23]. In OM, intra-operative samples were culture-positive in 52.65% (95% CI 30.54-74.22). Bacteraemia was detected in 23.91% (95% CI 8.40-44.24) of children with SA and 21.48% (95% CI 10.89-34.47) with OM. Despite appropriate sampling, a positive microbiological diagnosis is often lacking in paediatric acute osteoarticular infection using standard culture-based methods. This highlights the need for validation and use of more sensitive diagnostic methods, such as PCR.