A Data-Centric Approach To Generate Faithful and High Quality Patient Summaries with Large Language Models

Patients often face difficulties in understanding their hospitalizations, while healthcare workers have limited resources to provide explanations. In this work, we investigate the potential of large language models to generate patient summaries based on doctors' notes and study the effect of training data on the faithfulness and quality of the generated summaries. To this end, we develop a rigorous labeling protocol for hallucinations, and have two medical experts annotate 100 real-world summaries and 100 generated summaries. We show that fine-tuning on hallucination-free data effectively reduces hallucinations from 2.60 to 1.55 per summary for Llama 2, while preserving relevant information. Although the effect is still present, it is much smaller for GPT-4 when prompted with five examples (0.70 to 0.40). We also conduct a qualitative evaluation using hallucination-free and improved training data. GPT-4 shows very good results even in the zero-shot setting. We find that common quantitative metrics do not correlate well with faithfulness and quality. Finally, we test GPT-4 for automatic hallucination detection, which yields promising results.

Machine Learning for Health symposium 2023 -- Findings track

A collection of the accepted Findings papers that were presented at the 3rd Machine Learning for Health symposium (ML4H 2023), which was held on December 10, 2023, in New Orleans, Louisiana, USA. ML4H 2023 invited high-quality submissions on relevant problems in a variety of health-related disciplines including healthcare, biomedicine, and public health. Two submission tracks were offered: the archival Proceedings track, and the non-archival Findings track. Proceedings were targeted at mature work with strong technical sophistication and a high impact to health. The Findings track looked for new ideas that could spark insightful discussion, serve as valuable resources for the community, or could enable new collaborations. Submissions to the Proceedings track, if not accepted, were automatically considered for the Findings track. All the manuscripts submitted to ML4H Symposium underwent a double-blind peer-review process.

TabLLM: Few-shot Classification of Tabular Data with Large Language Models

We study the application of large language models to zero-shot and few-shot classification of tabular data. We prompt the large language model with a serialization of the tabular data to a natural-language string, together with a short description of the classification problem. In the few-shot setting, we fine-tune the large language model using some labeled examples. We evaluate several serialization methods including templates, table-to-text models, and large language models. Despite its simplicity, we find that this technique outperforms prior deep-learning-based tabular classification methods on several benchmark datasets. In most cases, even zero-shot classification obtains non-trivial performance, illustrating the method's ability to exploit prior knowledge encoded in large language models. Unlike many deep learning methods for tabular datasets, this approach is also competitive with strong traditional baselines like gradient-boosted trees, especially in the very-few-shot setting.

Large Language Models are Zero-Shot Clinical Information Extractors

We show that large language models, such as GPT-3, perform well at zero-shot information extraction from clinical text despite not being trained specifically for the clinical domain. We present several examples showing how to use these models as tools for the diverse tasks of (i) concept disambiguation, (ii) evidence extraction, (iii) coreference resolution, and (iv) concept extraction, all on clinical text. The key to good performance is the use of simple task-specific programs that map from the language model outputs to the label space of the task. We refer to these programs as resolvers, a generalization of the verbalizer, which defines a mapping between output tokens and a discrete label space. We show in our examples that good resolvers share common components (e.g., "safety checks" that ensure the language model outputs faithfully match the input data), and that the common patterns across tasks make resolvers lightweight and easy to create. To better evaluate these systems, we also introduce two new datasets for benchmarking zero-shot clinical information extraction based on manual relabeling of the CASI dataset (Moon et al., 2014) with labels for new tasks. On the clinical extraction tasks we studied, the GPT-3 + resolver systems significantly outperform existing zero- and few-shot baselines.