# DILA: Dictionary Label Attention for Mechanistic Interpretability in High-dimensional Multi-label Medical Coding Prediction

arXiv: https://arxiv.org/abs/2409.10504 · HTML: https://mlsanity.com/paper/2409.10504/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Interpreting biomedical and science models, one of 13 directions in Mechanistic interpretability.
- Builds on 22 of its 50 references that are in the map.
- First posted on arXiv 16 Sep 2024.

## Placed in

- Mechanistic interpretability / [Interpreting biomedical and science models](https://mlsanity.com/mechinterp/interpreting-biomedical-and-science-models/)

## Builds on (most cited first, 22 of 22)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [A Unified Approach to Interpreting Model Predictions](https://mlsanity.com/paper/1705.07874/) (2017)
- [Learning Important Features Through Propagating Activation Differences](https://mlsanity.com/paper/1704.02685/) (2017)
- [Toy Models of Superposition](https://mlsanity.com/paper/2209.10652/) (2022)
- [ERASER: A Benchmark to Evaluate Rationalized NLP Models](https://mlsanity.com/paper/1911.03429/) (2019)
- [Is Attention Interpretable?](https://mlsanity.com/paper/1906.03731/) (2019)
- [A Survey on Neural Network Interpretability](https://mlsanity.com/paper/2012.14261/) (2020)
- [Explainable Prediction of Medical Codes from Clinical Text](https://mlsanity.com/paper/1802.05695/) (2018)
- [One Explanation Does Not Fit All: A Toolkit and Taxonomy of AI Explainability Techniques](https://mlsanity.com/paper/1909.03012/) (2019)
- [An Attentive Survey of Attention Models](https://mlsanity.com/paper/1904.02874/) (2019)
- [Can Large Language Models Explain Themselves? A Study of LLM-Generated Self-Explanations](https://mlsanity.com/paper/2310.11207/) (2023)
- [Leveraging Sparse Linear Layers for Debuggable Deep Networks](https://mlsanity.com/paper/2105.04857/) (2021)
- [ICD Coding from Clinical Text Using Multi-Filter Residual Convolutional Neural Network](https://mlsanity.com/paper/1912.00862/) (2019)
- [A survey of sparse representation: algorithms and applications](https://mlsanity.com/paper/1602.07017/) (2016)
- [SPINE: SParse Interpretable Neural Embeddings](https://mlsanity.com/paper/1711.08792/) (2017)
- [Automated Medical Coding on MIMIC-III and MIMIC-IV: A Critical Review and Replicability Study](https://mlsanity.com/paper/2304.10909/) (2023)
- [A Comparative Study of Faithfulness Metrics for Model Interpretability Methods](https://mlsanity.com/paper/2204.05514/) (2022)
- [Explaining a Series of Models by Propagating Shapley Values](https://mlsanity.com/paper/2105.00108/) (2021)
- [White-Box Transformers via Sparse Rate Reduction: Compression Is All There Is?](https://mlsanity.com/paper/2311.13110/) (2023)
- [Surpassing GPT-4 Medical Coding with a Two-Stage Approach](https://mlsanity.com/paper/2311.13735/) (2023)
- [There is no Accuracy-Interpretability Tradeoff in Reinforcement Learning for Mazes](https://mlsanity.com/paper/2206.04266/) (2022)
- [Interpretation of multi-label classification models using shapley values](https://mlsanity.com/paper/2104.10505/) (2021)

## Built on by (most cited first, 4 of 4)

- [Make Mechanistic Interpretability Auditable: A Call to Develop Guidelines via Continuous Collaborative Reviewing](https://mlsanity.com/paper/2606.00033/) (2026)
- [Don't Pay Attention, PLANT It: Pretraining Attention via Learning-to-Rank](https://mlsanity.com/paper/2410.23066/) (2024)
- [Medical Interpretability and Knowledge Maps of Large Language Models](https://mlsanity.com/paper/2510.11390/) (2025)
- [Sparse Autoencoders for Mechanistic Interpretability in NLP: A Critical Survey of Recoverability, Interpretability, and Utility](https://mlsanity.com/paper/2610.20139/) (2026)

## Abstract (arXiv)

Predicting high-dimensional or extreme multilabels, such as in medical coding, requires both accuracy and interpretability. Existing works often rely on local interpretability methods, failing to provide comprehensive explanations of the overall mechanism behind each label prediction within a multilabel set. We propose a mechanistic interpretability module called DIctionary Label Attention (method) that disentangles uninterpretable dense embeddings into a sparse embedding space, where each nonzero element (a dictionary feature) represents a globally learned medical concept. Through human evaluations, we show that our sparse embeddings are more human understandable than its dense counterparts by at least 50 percent. Our automated dictionary feature identification pipeline, leveraging large language models (LLMs), uncovers thousands of learned medical concepts by examining and summarizing the highest activating tokens for each dictionary feature. We represent the relationships between dictionary features and medical codes through a sparse interpretable matrix, enhancing the mechanistic and global understanding of the model's predictions while maintaining competitive performance and scalability without extensive human annotation.