A Natural Language Processing and Machine Learning Approach to Identification of Incidental Radiology Findings in Trauma Patients Discharged from the Emergency Department.

STUDY OBJECTIVE: Patients undergoing diagnostic imaging studies in the emergency department (ED) commonly have incidental findings, which may represent unrecognized serious medical conditions, including cancer. Recognition of incidental findings frequently relies on manual review of textual radiology reports and can be overlooked in a busy clinical environment. Our study aimed to develop and validate a supervised machine learning model using natural language processing to automate the recognition of incidental findings in radiology reports of patients discharged from the ED. METHODS: We performed a retrospective analysis of computed tomography (CT) reports from trauma patients discharged home across an integrated health system in 2019. Two independent annotators manually labeled CT reports for the presence of an incidental finding as a reference standard. We used regular expressions to derive and validate a random forest model using open-source and machine learning software. Final model performance was assessed across different ED types. RESULTS: The study CT reports were divided into derivation (690 reports) and validation (282 reports) sets, with a prevalence of incidental findings of 22.3%, and 22.7%, respectively. The random forest model had an area under the curve of 0.88 (95% confidence interval [CI], 0.84 to 0.92) on the derivation set and 0.92 (95% CI, 0.88 to 0.96) on the validation set. The final model was found to have a sensitivity of 92.2%, a specificity of 79.4%, and a negative predictive value of 97.2%. Similarly, strong model performance was found when stratified to a dedicated trauma center, high-volume, and low-volume community EDs. CONCLUSION: Machine learning and natural language processing can classify incidental findings in CT reports of ED patients with high sensitivity and high negative predictive value across a broad range of ED settings. These findings suggest the utility of natural language processing in automating the review of free-text reports to identify incidental findings and may facilitate interventions to improve timely follow-up.

Mental health of the nursing and midwifery workforce in Wales during the COVID-19 pandemic: a cross-sectional analysis.

BACKGROUND: The mental health of the nursing and midwifery workforce in the UK became a public health concern before the COVID-19 pandemic. Poor mental health is a known factor for those considering leaving the profession, and workforce retention of younger members is crucial for the future of the sector. The aim of this study was to provide up-to-date estimates of mental wellbeing in this workforce in Wales during the COVID-19 pandemic. METHODS: We did a cross-sectional analysis of demographics, work-related information, and health data from respondents to a national online survey of registered and student nurses and midwives and health-care support workers in Wales. The survey was open between June 23 and Aug 9, 2021, and 2910 people responded (approximately 7% of the workforce). Mental wellbeing was calculated using the Short Warwick Edinburgh Mental Wellbeing Score (SWEMWBS). We measured probable clinical depression (SWEMWBS <18) and possible mild depression (SWEMWBS 18-20). We used χ2 analysis and multinomial logistic regression (adjusted for sex and staff grouping) to examine associations between age groups and mental wellbeing. FINDINGS: We analysed data from 2781 (95·6%) of 2910 respondents (129 respondents did not answer all seven SWEMWBS questions). Overall, 1622 (58·3%) of 2781 respondents had SWEMWBSs indicative of either probable clinical depression (863 [31·0%] of 2781) or possible mild depression (759 [27·3%] of 2781). Probable clinical depression was highest among those aged 18-29 years (180 [33·8%] of 532), 30-39 years (250 [35·6%] of 703), and 40-49 years (233 [33·5%] of 696). Respondents in these age groups were twice as likely to report SWEMWBSs indicative of probable clinical depression than respondents aged 60 years and older (18-29 years adjusted odds ratio [aOR] 2·38 [95% CI 1·43-3·97], p=0·0009; 30-39 years aOR 2·86 [1·77-4·64], p<0·0001; 40-49 years aOR 2·49 [1·54-4·02], p=0·0002). INTERPRETATION: This study highlights the substantial burden of poor mental wellbeing among the nursing and midwifery workforce in Wales, especially in those aged 49 years and younger. These figures, higher than previous estimates, could reflect the mental health effect of responding to the pandemic and could have long-term implications on workforce retention. FUNDING: None.

Factors influencing initiation of health behaviour conversations with patients: Cross-sectional study of nurses, midwives, and healthcare support workers in Wales.

AIM: To identify factors influencing healthcare professionals' engagement in health behaviour conversations with patients. DESIGN: Cross-sectional survey. METHODS: Between April and June 2019, an online survey of 1338 nurses, midwives and healthcare support workers was conducted. The survey assessed whether staff felt comfortable initiating health behaviour conversations with patients about five behaviours (reducing alcohol intake; stop smoking; being more active; reducing their weight; and improving their diet) and barriers to conversation initiation. Health professionals' own health-related behaviours, self-rated health and mental wellbeing, and socio-demographic characteristics were recorded. Logistic regression models were built to assess factors associated with feeling comfortable initiating health behaviour conversations for each topic. RESULT: Less than 50% of respondents reported feeling comfortable initiating health behaviour conversations with patients. Female staff, young professionals (18 to 29 years), those in lower staff grades and those with poorer health and low mental wellbeing were less likely to report feeling comfortable having health behaviour conversations across all topics. Those who did not adhere to physical activity and dietary guidelines were less likely to initiate a conversation about being more active and having a healthy diet, respectively. Not having time to discuss the topic, suitable space to hold a conversation, and feeling worried about offending/upsetting patients were the main barriers reported. CONCLUSION: Around 6 in 10 members of the nursing, midwifery and healthcare support workforce in Wales potentially do not feel comfortable to initiate a health behaviour conversation with patients about health and wellbeing. Feeling less comfortable to initiate a conversation was associated with staff demographics and organizational factors. IMPACT: We identified those less likely to initiate health behaviour conversations as well as personal and organizational barriers to initiation. This will help to target and tailor interventions to ensure staff are equipped and enabled to hold health behaviour conversations with patients.