{"id":18553917,"url":"https://github.com/soda-inria/predictive-ehr-benchmark","last_synced_at":"2025-05-15T12:07:30.508Z","repository":{"id":194579103,"uuid":"691136435","full_name":"soda-inria/predictive-ehr-benchmark","owner":"soda-inria","description":"Exploring a complexity gradient in representation and predictive algorithms for EHRs","archived":false,"fork":false,"pushed_at":"2023-09-14T11:59:16.000Z","size":79681,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":7,"default_branch":"main","last_synced_at":"2025-02-17T11:13:04.853Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"https://soda-inria.github.io/predictive-ehr-benchmark/","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/soda-inria.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null}},"created_at":"2023-09-13T15:12:28.000Z","updated_at":"2023-09-14T11:58:08.000Z","dependencies_parsed_at":"2023-09-14T06:08:51.942Z","dependency_job_id":null,"html_url":"https://github.com/soda-inria/predictive-ehr-benchmark","commit_stats":null,"previous_names":["soda-inria/predictive-ehr-benchmark"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/soda-inria%2Fpredictive-ehr-benchmark","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/soda-inria%2Fpredictive-ehr-benchmark/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/soda-inria%2Fpredictive-ehr-benchmark/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/soda-inria%2Fpredictive-ehr-benchmark/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/soda-inria","download_url":"https://codeload.github.com/soda-inria/predictive-ehr-benchmark/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":254337606,"owners_count":22054254,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-11-06T21:18:50.698Z","updated_at":"2025-05-15T12:07:25.496Z","avatar_url":"https://github.com/soda-inria.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Predictive algorithms from Electronic Health Records \n\nThis repository hosts code for the working paper: *Exploring a complexity gradient in representation and predictive algorithms for EHRs* \n\n[**Documentation**](https://soda-inria.github.io/predictive-ehr-benchmark/)\n\n[**Source Code**](https://github.com/soda-inria/predictive-ehr-benchmark)\n\n[**Working Paper repository**](https://github.com/strayMat/predictive_ehr_paper)\n\n### Abstract\n\nElectronic Health Records contain time-varying features with high cardinality.\nCurrent state-of-the-art predictive models build on increasingly elaborated\npipelines --based on transformers-- to handle the complexity of these data.\nAcknowledging the complexity to deploy, transfer and adapt these models on local\ncare environments, we explore a complexity-benefit tradeoff by comparing them to\nsimple aggregation of events. We use three clinical tasks involving time-varying\nstructured Electronic Health Records (EHRs) and increasingly clinically relevant\nproblems. We show that these benchmarking tasks display heterogeneous predictive\ndifficulties. We introduce a simple aggregation of static embeddings\n--transferred from national claims and publicly available--, showing that it\noutperforms transformer-based models on simple tasks with medium sample sizes.\nWe highlight the sample and computing resource efficiency of these models.\nFinally, clinically relevant problems generally present a strong class\nimbalance, which complicates models development and undermines their\nperformances. Further work is needed to understand if transformer-based models\nperform well in these scenarios where the number of cases requires good sample\nefficiency.\n\n# Usage\n\nSee the [usage page on the documentation](https://soda-inria.github.io/predictive-ehr-benchmark//usage.html)","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsoda-inria%2Fpredictive-ehr-benchmark","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fsoda-inria%2Fpredictive-ehr-benchmark","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsoda-inria%2Fpredictive-ehr-benchmark/lists"}