{
  "abstract": "Fluid and vasopressor management of sepsis requires a complex set of sequential treatment decisions tailored to individual responses. We examined the frequency and implications of disagreement between reinforcement-learning (RL), using the AI Clinician as a test example, and actual clinical decisions. A novel discordance score was defined as the Euclidean distance between clinical team actions and RL-recommended treatments (range 0 to 5.66), across 25 treatment options of vasopressors and intravenous (IV) fluids. Among 30,678 ICU patients with sepsis cared for across 309,415 time-epochs, discordance was more common among epochs of patients with SOFA > 6 than those with SOFA < 6, and for intravenous fluid choices than for vasopressors. After multivariable adjustment, greater discordance was associated with 90-day mortality (adjusted odds ratio (OR) 1.11, 95%CI: 1.05,1.18) in the SOFA strata > =6 but not for SOFA < = 4 (adjusted odds ratio (OR): 1.02, 95%CI: 0.97,1.08).",
  "abstract_shown": true,
  "affiliations": [
    "Division of Pulmonary, Allergy, Critical Care, and Sleep Medicine, Department of Medicine, University of Pittsburgh Medical Center",
    "Department of Critical Care Medicine, University of Pittsburgh School of Medicine",
    "Department of Biostatistics, University of Pittsburgh",
    "Department of Surgery and Cancer, Imperial College London",
    "Department of Anesthesia and Perioperative Care, Zuckerberg San Francisco General Hospital and Trauma Center, University of California San Francisco",
    "Department of Computer Science, University of Pittsburgh"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 11,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "Peter C. Nauka"
    },
    {
      "affs": [
        1
      ],
      "name": "Jason N. Kennedy"
    },
    {
      "affs": [
        2
      ],
      "name": "Xinlei Chen"
    },
    {
      "affs": [
        3
      ],
      "name": "Matthieu Komorowski"
    },
    {
      "affs": [
        2
      ],
      "name": "Chung-Chou H. Chang"
    },
    {
      "affs": [
        4
      ],
      "name": "Romain Pirracchio"
    },
    {
      "affs": [
        2
      ],
      "name": "Lu Tang"
    },
    {
      "affs": [
        1
      ],
      "name": "Bryan J. McVerry"
    },
    {
      "affs": [
        1
      ],
      "name": "Derek C. Angus"
    },
    {
      "affs": [
        5
      ],
      "name": "Milos Hauskrecht"
    },
    {
      "affs": [
        1
      ],
      "name": "Christopher W. Seymour"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/8a5bm4698mc3w.bib",
  "cited_by": 0,
  "doi": "10.1038/s44401-026-00156-2",
  "fulltext_url": "https://doi.org/10.1038/s44401-026-00156-2",
  "id": "8a5bm4698mc3w",
  "issns": [
    "3005-1959"
  ],
  "landing_url": "https://doi.org/10.1038/s44401-026-00156-2",
  "language": "en",
  "licence": "CC-BY",
  "licence_label": "CC BY",
  "licence_source": "journal",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "lcc",
      "RA1-1270"
    ],
    [
      "openalex",
      "2713|Sepsis Diagnosis and Treatment"
    ]
  ],
  "openalex": "W7223174488",
  "pdf_url": "https://www.nature.com/articles/s44401-026-00156-2.pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-25",
  "publisher": "Nature Publishing Group",
  "ris": "https://research.draconitic.com/paper/8a5bm4698mc3w.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T13:45:35Z",
      "native_id": "87f07b8a103e400a8a1a0709421a7cff",
      "source": "doaj",
      "url": "https://doi.org/10.1038/s44401-026-00156-2"
    }
  ],
  "subjects": [
    {
      "field": "medicine",
      "origin": "openalex:2713|Sepsis Diagnosis and Treatment",
      "subfield": "epidemiology-public-health"
    }
  ],
  "title": "Discordance between actual care and recommendations by a reinforcement learning model for sepsis",
  "title_text": "Discordance between actual care and recommendations by a reinforcement learning model for sepsis",
  "type": "peer-reviewed",
  "updated_at": "2026-10-30T13:45:35Z",
  "url": "https://research.draconitic.com/paper/8a5bm4698mc3w",
  "venue": "npj Health Systems",
  "versions": []
}