Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
194 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Inferring Actual Treatment Pathways from Patient Records (2309.01897v3)

Published 5 Sep 2023 in cs.LG and cs.AI

Abstract: Treatment pathways are step-by-step plans outlining the recommended medical care for specific diseases; they get revised when different treatments are found to improve patient outcomes. Examining health records is an important part of this revision process, but inferring patients' actual treatments from health data is challenging due to complex event-coding schemes and the absence of pathway-related annotations. This study aims to infer the actual treatment steps for a particular patient group from administrative health records (AHR) - a common form of tabular healthcare data - and address several technique- and methodology-based gaps in treatment pathway-inference research. We introduce Defrag, a method for examining AHRs to infer the real-world treatment steps for a particular patient group. Defrag learns the semantic and temporal meaning of healthcare event sequences, allowing it to reliably infer treatment steps from complex healthcare data. To our knowledge, Defrag is the first pathway-inference method to utilise a neural network (NN), an approach made possible by a novel, self-supervised learning objective. We also developed a testing and validation framework for pathway inference, which we use to characterise and evaluate Defrag's pathway inference ability and compare against baselines. We demonstrate Defrag's effectiveness by identifying best-practice pathway fragments for breast cancer, lung cancer, and melanoma in public healthcare records. Additionally, we use synthetic data experiments to demonstrate the characteristics of the Defrag method, and to compare Defrag to several baselines where it significantly outperforms non-NN-based methods. Defrag significantly outperforms several existing pathway-inference methods and offers an innovative and effective approach for inferring treatment pathways from AHRs. Open-source code is provided to encourage further research in this area.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (56)
  1. Process mining. Communications of the ACM 55, 76–83. URL: https://doi.org/10.1145%2F2240236.2240257, doi:10.1145/2240236.2240257.
  2. Topology of evolving networks: local events and universality. Physical review letters 85, 5234.
  3. Process mining routinely collected electronic health records to define real-life clinical pathways during chemotherapy. International Journal of Medical Informatics 103, 32–41. URL: https://doi.org/10.1016%2Fj.ijmedinf.2017.03.011, doi:10.1016/j.ijmedinf.2017.03.011.
  4. A dendrite method for cluster analysis. Communications in Statistics - Theory and Methods 3, 1–27. URL: https://doi.org/10.1080%2F03610927408827101, doi:10.1080/03610927408827101.
  5. Early breast cancer: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Annals of Oncology 30, 1194–1220. URL: https://doi.org/10.1093%2Fannonc%2Fmdz173, doi:10.1093/annonc/mdz173.
  6. Textual analysis and visualization of research trends in data mining for electronic health records. Health Policy and Technology 6, 389–400. URL: https://doi.org/10.1016%2Fj.hlpt.2017.10.003, doi:10.1016/j.hlpt.2017.10.003.
  7. Multi-layer representation learning for medical concepts, ACM. URL: https://doi.org/10.1145%2F2939672.2939823, doi:10.1145/2939672.2939823.
  8. Mime: Multilevel medical embedding of electronic health records for predictive healthcare, in: 32nd Conference on Neural Information Processing Systems, NeurIPS 2018, pp. 4547–4557.
  9. A cluster separation measure. IEEE Transactions on Pattern Analysis and Machine Intelligence PAMI-1, 224–227. URL: https://doi.org/10.1109%2Ftpami.1979.4766909, doi:10.1109/tpami.1979.4766909.
  10. Getting a grasp on clinical pathway data: An approach based on process mining, in: Washio, T., Luo, J. (Eds.), Emerging Trends in Knowledge Discovery and Data Mining. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 22–35. doi:10.1007/978-3-642-36778-6_3.
  11. Variations in patients’ adherence to medical recommendations. Medical Care 42, 200–209. URL: https://doi.org/10.1097%2F01.mlr.0000114908.90348.f9, doi:10.1097/01.mlr.0000114908.90348.f9.
  12. Adherence to guidelines and protocols in the prehospital and emergency care setting: a systematic review. Scandinavian Journal of Trauma, Resuscitation and Emergency Medicine 21. URL: https://doi.org/10.1186%2F1757-7241-21-9, doi:10.1186/1757-7241-21-9.
  13. How do physicians use practice guidelines? Drug Benefit Trends 19, 237.
  14. Simcse: Simple contrastive learning of sentence embeddings. arXiv preprint arXiv:2104.08821 .
  15. An introduction to hidden markov models and bayesian networks, in: Series in Machine Perception and Artificial Intelligence. World Scientific, pp. 9–41. URL: https://doi.org/10.1142%2F9789812797605_0002, doi:10.1142/9789812797605_0002.
  16. Generation and evaluation of synthetic patient data. BMC Medical Research Methodology 20. URL: https://doi.org/10.1186%2Fs12874-020-00977-1, doi:10.1186/s12874-020-00977-1.
  17. Process mining applications in the healthcare domain: A comprehensive review. WIREs Data Mining and Knowledge Discovery 12. URL: https://doi.org/10.1002%2Fwidm.1442, doi:10.1002/widm.1442.
  18. Music transformer. arXiv preprint URL: https://arxiv.org/abs/1809.04281, arXiv:1809.04281.
  19. Discovery of clinical pathway patterns from event logs using probabilistic topic models. Journal of Biomedical Informatics 47, 39--57. URL: https://doi.org/10.1016%2Fj.jbi.2013.09.003, doi:10.1016/j.jbi.2013.09.003.
  20. Probabilistic modeling personalized treatment pathways using electronic health records. Journal of Biomedical Informatics 86, 33--48. URL: https://doi.org/10.1016%2Fj.jbi.2018.08.004, doi:10.1016/j.jbi.2018.08.004.
  21. Latent treatment pattern discovery for clinical processes. Journal of Medical Systems 37. URL: https://doi.org/10.1007%2Fs10916-012-9915-2, doi:10.1007/s10916-012-9915-2.
  22. Mimic-iv (version 0.4). PhysioNet .
  23. Process mining for clinical workflows: challenges and current limitations, in: EHealth beyond the horizon: get it there: proceedings of MIE2008 the XXIst international congress of the european federation for medical informatics, p. 229.
  24. Deriving a sophisticated clinical pathway based on patient conditions from electronic health record data, in: Leemans, S., Leopold, H. (Eds.), Lecture Notes in Business Information Processing. Springer International Publishing, pp. 356--367. URL: https://doi.org/10.1007%2F978-3-030-72693-5_27, doi:10.1007/978-3-030-72693-5_27.
  25. Assessment of the feasibility of developing a clinical pathway using a clinical order log. Journal of Biomedical Informatics 128, 104038. URL: https://doi.org/10.1016%2Fj.jbi.2022.104038, doi:10.1016/j.jbi.2022.104038.
  26. Can process mining automatically describe care pathways of patients with long-term conditions in UK primary care? a study protocol. BMJ Open 8, e019947. URL: https://doi.org/10.1136%2Fbmjopen-2017-019947, doi:10.1136/bmjopen-2017-019947.
  27. Decoupled weight decay regularization. arXiv preprint URL: https://arxiv.org/abs/1711.05101, arXiv:1711.05101.
  28. HDBSCAN: Hierarchical density based clustering. The Open Journal 2, 205. URL: https://doi.org/10.21105/joss.00205, doi:10.21105/joss.00205.
  29. Cutaneous melanoma: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Annals of Oncology 30, 1884--1901. URL: https://doi.org/10.1093%2Fannonc%2Fmdz411, doi:10.1093/annonc/mdz411.
  30. Distributed representations of words and phrases and their compositionality. Advances in neural information processing systems 26.
  31. Are we ready for conformance checking in healthcare? measuring adherence to clinical guidelines: A scoping systematic literature review. Journal of Biomedical Informatics 130, 104076. URL: https://doi.org/10.1016%2Fj.jbi.2022.104076, doi:10.1016/j.jbi.2022.104076.
  32. Reducing clinical variations with clinical pathways: do pathways work? International Journal for Quality in Health Care 15, 509--521. URL: https://doi.org/10.1093%2Fintqhc%2Fmzg057, doi:10.1093/intqhc/mzg057.
  33. Zipf’s word frequency law in natural language: A critical review and future directions. Psychonomic Bulletin & Review 21, 1112--1130. URL: https://doi.org/10.3758%2Fs13423-014-0585-6, doi:10.3758/s13423-014-0585-6.
  34. Business process analysis in healthcare environments: A methodology based on process mining. Information Systems 37, 99--116. URL: https://doi.org/10.1016%2Fj.is.2011.01.003, doi:10.1016/j.is.2011.01.003.
  35. The effects of clinical pathways on professional practice, patient outcomes, length of stay, and hospital costs. Evaluation & the Health Professions 35, 3--27. URL: https://doi.org/10.1177%2F0163278711407313, doi:10.1177/0163278711407313.
  36. Clinical pathways: effects on professional practice, patient outcomes, length of stay and hospital costs. Cochrane Database of Systematic Reviews URL: https://doi.org/10.1002%2F14651858.cd006632.pub2, doi:10.1002/14651858.cd006632.pub2.
  37. Silhouettes: A graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied Mathematics 20, 53--65. URL: https://doi.org/10.1016%2F0377-0427%2887%2990125-7, doi:10.1016/0377-0427(87)90125-7.
  38. A distance measure between attributed relational graphs for pattern recognition. IEEE Transactions on Systems, Man, and Cybernetics SMC-13, 353--362. URL: https://doi.org/10.1109%2Ftsmc.1983.6313167, doi:10.1109/tsmc.1983.6313167.
  39. Self-attention with relative position representations. arXiv preprint URL: https://arxiv.org/abs/1803.02155, arXiv:1803.02155.
  40. Weisfeiler-lehman graph kernels. Journal of Machine Learning Research 12.
  41. 2nd ESMO consensus conference on lung cancer: early-stage non-small-cell lung cancer consensus on diagnosis, treatment and follow-up. Annals of Oncology 25, 1462--1474. URL: https://doi.org/10.1093%2Fannonc%2Fmdu089, doi:10.1093/annonc/mdu089.
  42. Attention is all you need, in: Proceedings of the 31st International Conference on Neural Information Processing Systems, pp. 5998--6008.
  43. Information theoretic measures for clusterings comparison, ACM Press. URL: https://doi.org/10.1145%2F1553374.1553511, doi:10.1145/1553374.1553511.
  44. Individual comparisons by ranking methods. Biometrics Bulletin 1, 80. URL: https://doi.org/10.2307%2F3001968, doi:10.2307/3001968.
  45. A New Kind of Science. Wolfram Media. URL: https://www.wolframscience.com.
  46. TCPM: Topic-based clinical pathway mining, in: 2016 IEEE First International Conference on Connected Health: Applications, Systems and Engineering Technologies (CHASE), IEEE. URL: https://doi.org/10.1109%2Fchase.2016.17, doi:10.1109/chase.2016.17.
  47. Incorporating topic assignment constraint and topic correlation limitation into clinical goal discovering for clinical pathway mining. Journal of Healthcare Engineering 2017, 1--13. URL: https://doi.org/10.1155%2F2017%2F5208072, doi:10.1155/2017/5208072.
  48. Mining electronic health records (EHRs). ACM Computing Surveys 50, 1--40. URL: https://doi.org/10.1145%2F3127881, doi:10.1145/3127881.
  49. Learning clinical workflows to identify subgroups of heart failure patients, in: AMIA Annual Symposium Proceedings, American Medical Informatics Association. p. 1248.
  50. Process mining for clinical pathway: Literature review and future directions, in: 2014 11th International Conference on Service Systems and Service Management (ICSSSM), IEEE. URL: https://doi.org/10.1109%2Ficsssm.2014.6943412, doi:10.1109/icsssm.2014.6943412.
  51. Combining deep learning with token selection for patient phenotyping from electronic health records. Scientific Reports 10. URL: https://doi.org/10.1038%2Fs41598-020-58178-1, doi:10.1038/s41598-020-58178-1.
  52. Electronic health records (EHRs): Supporting ASCO's vision of cancer care. American Society of Clinical Oncology Educational Book , 225--231URL: https://doi.org/10.14694%2Fedbook_am.2014.34.225, doi:10.14694/edbook_am.2014.34.225.
  53. Barlow twins: Self-supervised learning via redundancy reduction, in: International Conference on Machine Learning, PMLR. pp. 12310--12320.
  54. Inferring ehr utilization workflows through audit logs, in: AMIA Annual Symposium Proceedings, American Medical Informatics Association. p. 1247.
  55. Paving the COWpath: Learning and visualizing clinical pathways from electronic health record data. Journal of Biomedical Informatics 58, 186--197. URL: https://doi.org/10.1016%2Fj.jbi.2015.09.009, doi:10.1016/j.jbi.2015.09.009.
  56. On clinical pathway discovery from electronic health record data. IEEE Intelligent Systems 30, 70--75. URL: https://doi.org/10.1109%2Fmis.2015.14, doi:10.1109/mis.2015.14.
Citations (2)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now