From dictations to clinical reports using machine translation

Abstract

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.

Towards Automating Medical Scribing : Clinic Visit Dialogue2Note Sentence Alignment and Snippet Summarization

“…Creating a corpus of aligned clinic visit conversation dialogue sentences with corresponding clinical note sentences is instrumental for training language generation systems. Early work in this domain includes that of (Finley et al, 2018), which uses an automated algorithm based on some heuristics, e.g. string matches, and merge conditions, to align dictation parts of clinical notes.…”

“…Clinical Language Generation from Conversation (Finley et al, 2018) produced dictation parts of a report, measuring performance both on gold standard transcripts and raw ASR output using statistical MT methods. In (Liu et al, 2019), the authors labeled a corpus of 101K simulated conversations and 490 nurse-patient dialogues with artificial short semi-structured summaries.…”

The Medical Scribe: Corpus Development and Model Performance Analyses

“…One possible explanation is that unlike entities, the span boundaries of attributes may be less distinct in a conversation. We investigated this by relaxing the model requirements, from identifying the span of words to detecting speaker turns containing the attributes, which happens to be similar to other previous work (Finley et al, 2018b;Lacson et al, 2006;Park et al, 2019).…”

“…This seems to perform well in a narrowly scoped task. A more ambitious approach mapped ASR transcripts to clinical notes by adopting a machine translation approach (Finley et al, 2018b). However this performed poorly.…”

Challenges of developing a digital scribe to reduce clinical documentation burden

“…To date, research effort has focused on solving foundational problems in the development of a digital scribe, including ASR of medical conversations, 10,11 automatically populating the review of symptoms discussed in a medical encounter, 12 extracting symptoms from medical conversations, 13,14 and generating medical reports from dictations. 15,16 While these developments are promising, several challenges hinder the implementation of a fully functioning digital scribe and its evaluation in a clinical environment. This paper will discuss the major challenges, with a summary presented in Table 1.…”