Abstract:Generating clinical reports from raw recordings such as X-rays and electroencephalogram (EEG) is an essential and routine task for doctors. However, it is often time-consuming to write accurate and detailed reports. Most existing methods try to generate the whole reports from the raw input with limited success because 1) generated reports often contain errors that need manual review and correction, 2) it does not save time when doctors want to write additional information into the report, and 3) the generated reports are not customized based on individual doctors' preference. We propose {\it CL}inic{\it A}l {\it R}eport {\it A}uto-completion (CLARA), an interactive method that generates reports in a sentence by sentence fashion based on doctors' anchor words and partially completed sentences. CLARA searches for most relevant sentences from existing reports as the template for the current report. The retrieved sentences are sequentially modified by combining with the input feature representations to create the final report. In our experimental evaluation, CLARA achieved 0.393 CIDEr and 0.248 BLEU-4 on X-ray reports and 0.482 CIDEr and 0.491 BLEU-4 for EEG reports for sentence-level generation, which is up to 35% improvement over the best baseline. Also via our qualitative evaluation, CLARA is shown to produce reports which have a significantly higher level of approval by doctors in a user study (3.74 out of 5 for CLARA vs 2.52 out of 5 for the baseline).

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is how to improve the efficiency and accuracy of clinical report generation. Specifically, the existing methods have the following limitations in generating complete reports: 1. **Errors in Generated Reports**: The generated reports often contain errors that require manual review and correction. 2. **Time - consuming for Doctors to Add Information**: When doctors want to add extra information to the reports, it does not save time. 3. **Lack of Personalized Customization**: The generated reports are not customized according to the preferences of individual doctors. To solve these problems, the authors propose an interactive method named CLARA (CLinicAl Report Auto - completion), which generates reports sentence by sentence based on the anchor words (disease phenotype keywords) provided by doctors and partially completed sentences. CLARA improves the existing methods in the following ways: - **Aligning Disease Phenotypes**: Ensure that the generated reports are aligned with disease phenotypes and use medical terms correctly. - **Adaptive Report Generation**: The generated reports can be adjusted according to the preferences of end - users (such as doctors) to improve the adoption rate. ### Specific Methods The CLARA framework includes the following modules: 1. **Input Encoder Module**: - Use a convolutional neural network (CNN) to extract compressed feature representations from X - ray images or EEG time series. For example, for X - ray images, use the DenseNet architecture for encoding; for EEG data, average the feature vectors of all epochs to obtain the final embedding. 2. **Prototype Construction Module**: - Construct a prototype library containing all unique sentences in previous reports and assign weights to each sentence. These prototype sentences and their representations are indexed for quick retrieval. 3. **Query Module**: - Accept the anchor words (global context) and prefix text (local context) provided by doctors to guide report generation. Anchor words are phenotype keywords related to the entire report, while prefix text is the partially entered sentence by the user. 4. **Retrieve and Edit Module**: - Retrieve the most relevant sentence templates from the prototype library through an information retrieval system (such as Lucene), and then use a sequence - to - sequence (seq2seq) model to modify these sentences according to the input data, anchor words, and prefix text. ### Experimental Results The authors evaluated the effectiveness of CLARA on two tasks: 1. **X - ray Report Generation**: Using the IU X - ray dataset, CLARA achieved a CIDEr score of 0.393 and a BLEU - 4 score of 0.248. 2. **EEG Report Generation**: Using the TUH EEG and MGH EEG datasets, CLARA achieved CIDEr scores of 0.482 and 0.491 respectively, and also significantly outperformed the baseline model in BLEU - 4 scores. In addition, user studies show that doctors' quality scores (3.74/5) for CLARA - generated reports are significantly higher than those of the baseline model (2.52/5). In conclusion, CLARA significantly improves the quality and efficiency of clinical report generation through interactive generation and customization functions.

CLARA: Clinical Report Auto-completion

An Inclusive Task-Aware Framework for Radiology Report Generation

Automated Generation of Clinical Reports Using Sensing Technologies with Deep Learning Techniques

Clinically Accurate Chest X-Ray Report Generation

Consensus, dissensus and synergy between clinicians and specialist foundation models in radiology report generation

Radiology Report Generation via Structured Knowledge-Enhanced Multi-modal Attention and Contrastive Learning.

Collaboration between clinicians and vision–language models in radiology report generation

MAIRA-2: Grounded Radiology Report Generation

On the Automatic Generation of Medical Imaging Reports

Automated Generation of Accurate \& Fluent Medical X-ray Reports

Evaluating Automated Radiology Report Quality through Fine-Grained Phrasal Grounding of Clinical Findings

AutoRG-Brain: Grounded Report Generation for Brain MRI

Agentic LLM Workflows for Generating Patient-Friendly Medical Reports

A medical report generation method integrating teacher–student model and encoder–decoder network

Patient-centered radiology reports with generative artificial intelligence: adding value to radiology reporting

RadioBERT: A deep learning-based system for medical report generation from chest X-ray images using contextual embeddings

An efficient but effective writer: Diffusion-based semi-autoregressive transformer for automated radiology report generation

Work like a doctor: Unifying scan localizer and dynamic generator for automated computed tomography report generation

Practical Evaluation of ChatGPT Performance for Radiology Report Generation

AI-Assisted Summarization of Radiologic Reports: Evaluating GPT3davinci, BARTcnn, LongT5booksum, LEDbooksum, LEDlegal, and LEDclinical

Improving Clinical Note Generation from Complex Doctor-Patient Conversation