{"paper":{"title":"Uncertainty-Driven Anomaly Detection for Psychotic Relapse Using Smartwatches: Forecasting and Multi-Task Learning Fusion","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Late fusion of cardiac forecasting and multi-task sleep-motion models on smartwatches detects psychotic relapse with an 8% improvement over the winning baseline.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Niki Efthymiou, Nikolaos Tsalkitzis, Panagiotis P.Filntisis, Petros Maragos","submitted_at":"2026-05-13T17:43:07Z","abstract_excerpt":"Digital phenotyping enables continuous passive monitoring of behavior and physiology, offering a promising paradigm for early detection of psychotic relapse. In this work, we develop and systematically study two smartwatch-based frameworks for daily relapse detection. The first forecasts cardiac dynamics and flags deviations between predicted and observed features as indicators of abnormality. The second adopts a multi-task formulation that fuses sleep with motion and cardiac-derived signals, learning time-aware embeddings and predicting measurement timing. Both pipelines use Transformer encod"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"our fused model achieves a 8% relative improvement over the competition-winning baseline... the integration of diverse digital phenotypes, cardiac, motion, and sleep, is essential for the high-fidelity detection of psychotic relapse in real-world settings.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That deviations flagged as anomalies by the uncertainty-driven scores correspond to actual clinical psychotic relapse events rather than other sources of wearable noise or behavioral change.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Fusing cardiac forecasting with multi-task sleep-motion learning on smartwatch data produces an 8% better psychotic relapse detector than the prior competition winner.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Late fusion of cardiac forecasting and multi-task sleep-motion models on smartwatches detects psychotic relapse with an 8% improvement over the winning baseline.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"51e05e49f079651bc7ba3826ebb4647720096e255d3a2f03aa23d5f7c5101126"},"source":{"id":"2605.13816","kind":"arxiv","version":1},"verdict":{"id":"36e28264-a4e7-45a0-ab1c-7e528504b007","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:16:16.824307Z","strongest_claim":"our fused model achieves a 8% relative improvement over the competition-winning baseline... the integration of diverse digital phenotypes, cardiac, motion, and sleep, is essential for the high-fidelity detection of psychotic relapse in real-world settings.","one_line_summary":"Fusing cardiac forecasting with multi-task sleep-motion learning on smartwatch data produces an 8% better psychotic relapse detector than the prior competition winner.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That deviations flagged as anomalies by the uncertainty-driven scores correspond to actual clinical psychotic relapse events rather than other sources of wearable noise or behavioral change.","pith_extraction_headline":"Late fusion of cardiac forecasting and multi-task sleep-motion models on smartwatches detects psychotic relapse with an 8% improvement over the winning baseline."},"references":{"count":19,"sample":[{"doi":"","year":2021,"title":"Opportunities and challenges in the collection and anal- ysis of digital phenotyping data,","work_id":"a5ac8e2f-2e1c-48ed-86c1-d7a5ce929e3b","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Digital phenotyping: a global tool for psychiatry,","work_id":"a4285cd9-1d98-43f3-a666-b5db7bcda484","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Automatic recognition of schizophrenia from facial videos using 3D convolutional neural net- work,","work_id":"553ecb3c-7732-4125-8fc9-473baf32304b","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Predicting early warning signs of psychotic relapse from passive sensing data: an approach using encoder-decoder neural networks,","work_id":"7b4ee2a2-b078-426e-99ba-3d4cbc99afe4","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Smartphone Health Assessment for Relapse Prevention (SHARP): a digital solution toward global mental health,","work_id":"35e24c18-9893-4873-a356-4629995b2c57","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":19,"snapshot_sha256":"b9a2a6e20293698be5bd1b0e05a95fdf413468ad86be349fa53202fd67e62b84","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}