A medical scribe is a clinical professional who charts patient-physician encounters in real time, relieving physicians of most of their administrative burden and substantially increasing productivity and job satisfaction. We present a complete implementation of an automated medical scribe. Our system can serve either as a scalable, standardized, and economical alternative to human scribes; or as an assistive tool for them, providing a first draft of a report along with a convenient means to modify it. This solution is, to our knowledge, the first automated scribe ever presented and relies upon multiple speech and language technologies, including speaker diarization, medical speech recognition, knowledge extraction, and natural language generation.

A typical workflow to document clinical encounters entails dictating a summary, running speech recognition, and post-processing the resulting text into a formatted letter. Post-processing entails a host of transformations including punctuation restoration, truecasing, marking sections and headers, converting dates and numerical expressions, parsing lists, etc. In conventional implementations, most of these tasks are accomplished by individual modules. We introduce a novel holistic approach to post-processing that relies on machine callytranslation. We show how this technique outperforms an alternative conventional system-even learning to correct speech recognition errors during post-processingwhile being much simpler to maintain.

No abstract

No abstract

Introduction Few studies have investigated oesophageal cancer care in regional areas. This study aimed to describe treatment patterns for oesophageal cancer in a regional area, and to identify factors associated with radiotherapy utilisation, timeliness of care, and death. Methods In a retrospective cohort study, medical records were reviewed to source data on all patients diagnosed with and/or treated for oesophageal cancer at two regional Victorian hospitals over July 2015–June 2018. Cox proportional hazards regression was employed to identify factors associated with time from diagnosis to death while binary logistic regression was used to identify factors associated with radiotherapy utilisation and treatment within 28 days of diagnosis – a time frame derived from the relevant optimal care pathway. Results Of 95 patients, 72% had radiotherapy, 32% received any treatment within 28 days, and 78% died over a median time of nine months. Odds of not receiving radiotherapy were decreased (odds ratio [OR] = 0.26, 95% confidence interval [CI] = 0.08–0.87) for histology other than adenocarcinoma. Odds of timely care were increased for any palliative radiotherapy (OR = 3.47, 95% CI = 1.15–10.5) and decreased for older age (OR = 0.95, 95% CI = 0.91.0.999). Hazard of death was elevated for stage IV disease (hazard ratio [HR] = 2.73, 95% CI = 1.64–4.54) and reduced for radical intent (HR = 0.27, 95% CI = 0.15–0.48). Conclusion Nearly three‐quarters of regional oesophageal cancer patients had radiotherapy while approximately one‐third received any treatment within the recommended 28 days. Any palliative radiotherapy and younger age were associated with timely treatment. Future studies could further investigate factors related to timely oesophageal cancer care.

No abstract

Background Timeliness of cancer care is vital for improved survival and quality of life of patients. Service and care centralisation at larger‐volume centres has been associated with improved outcomes. However, there is a lack of systematic data on the impact of tumour stream volume on timeliness of care. Aims To investigate and compare timeliness of care for lung cancer, a high‐volume (more commonly diagnosed) tumour stream, and oesophagogastric (OG) cancer, a low‐volume (less commonly diagnosed) tumour stream, at a regional health service in Victoria, Australia. Methods A retrospective cohort study comprising random samples of 75 people newly diagnosed with lung cancer (International Classification of Diseases and Related Health Problems‐10 [ICD‐10] diagnosis codes C34 in the Victorian Cancer Registry [VCR]) and 50 people newly diagnosed with OG cancer (ICD‐10 diagnosis codes C15 or C16 in VCR) at one regional Victorian health service between 2016 and 2017. Binary logistic regression was used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for associations between patient factors and suboptimal timeliness of care. Results In comparison to OG cancer patients, lung cancer patients had reduced odds of suboptimal timeliness of care in reference to times outside OCP for referral to diagnosis (OR [95% CI] = 0.34 [0.14 to 0.83]) but increased odds of suboptimal timeliness for diagnosis to treatment (OR [95% CI] = 2.48 [1.01 to 6.09]). Conclusion In the low‐volume OG cancer stream, patients had longer wait times from referral to an MDM, where treatment decisions occur, but shorter time to commencement of first treatment. Conversely in the high‐volume lung cancer group, there was delayed initiation of first treatment following presentation at MDM. There is need to explore ways to fast‐track MDM presentation and commencement of therapy among people diagnosed with low‐volume and high‐volume cancers, respectively.

No abstract