Madical AI Use — plain-text case export Not medical advice. Illustrative entries are composites, not trial reports. Source index: https://madicalaiuse.ai # Autonomous diabetic retinopathy screening in primary care slug: lumineticscore-dr-screening category: imaging-radiology evidence: regulatory source_type: news year: 2018 source: U.S. FDA De Novo DEN180001 source_url: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/denovo.cfm?id=DEN180001 source_date: 2018-04-11 tools: LumineticsCore (IDx-DR); Digital Diagnostics; Topcon NW400 tags: screening, regulation, computer-vision url: https://madicalaiuse.ai/en/cases/lumineticscore-dr-screening url_zh: https://madicalaiuse.ai/zh-CN/cases/lumineticscore-dr-screening IDx-DR, now marketed as LumineticsCore by Digital Diagnostics, was authorized to screen adults with diabetes for more-than-mild diabetic retinopathy without an eye specialist reading the photo first. A primary care team captures images on a specified retinal camera and the software returns a refer-or-rescreen result. It is not a complete eye examination. Pregnancy and several existing retinal conditions sit outside the labeled use. Setting: Primary care and endocrinology clinics that can take retinal photographs during a diabetes visit, with referral to eye care when the result is positive. Reported: In the study FDA reviewed (900 patients, 10 primary care sites), the system correctly identified more-than-mild diabetic retinopathy 87.4% of the time and correctly identified patients without it 89.5% of the time. ZH-CN title: 基层医疗中的糖尿病视网膜病变自主筛查 IDx-DR（现由 Digital Diagnostics 以 LumineticsCore 销售）获授权用于筛查成年糖尿病患者是否存在超过轻度的糖尿病视网膜病变，且无需眼科医生先读片。基层团队用指定眼底相机拍照，软件给出“转诊”或“复查”两类结果。它不能代替全面眼科检查。妊娠以及若干既有视网膜病变不在适应证内。 ---- # AI-supported reading in the MASAI mammography trial slug: masai-mammography-screening category: imaging-radiology evidence: published source_type: paper year: 2023 source: Lång et al., The Lancet Oncology source_url: https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(23)00298-X/fulltext source_date: 2023-08-01 tools: Transpara 1.7.0; ScreenPoint Medical tags: screening, computer-vision, worklist url: https://madicalaiuse.ai/en/cases/masai-mammography-screening url_zh: https://madicalaiuse.ai/zh-CN/cases/masai-mammography-screening The MASAI trial in Sweden randomly assigned screening mammograms to AI-supported reading or standard double reading without AI. Transpara scored each exam and routed lower scores to a single reader and the highest score to double reading, with detection marks available on higher-risk exams. This card quotes the prespecified interim safety analysis, not a claim that every screening program should switch. Setting: Population mammography screening for women aged 40–80 at four sites in southwest Sweden, inside the existing screening invitation. Reported: In the interim analysis of about 80,000 women (Lancet Oncology, 2023), cancer detection was 6.1 per 1,000 with AI support and 5.1 per 1,000 with standard reading (ratio 1.2, p=0.052). False-positive rates were 1.5% in both arms. Screen-reading workload fell by 44.3%. Interval cancer, the primary endpoint, was left to later follow-up. ZH-CN title: MASAI 试验中的人工智能辅助乳腺筛查读片 瑞典 MASAI 试验将筛查钼靶随机分到人工智能辅助读片，或无人工智能的常规双人读片。Transpara 为每次检查评分：较低分数由单人阅读，最高分数进入双人阅读，较高风险检查可显示检出标记。本条只转述预先设定的中期安全性分析，并不表示所有筛查项目都应切换。 ---- # Worklist triage for critical findings on chest radiographs slug: chest-xray-critical-triage category: imaging-radiology evidence: vendor-reported source_type: news year: 2024 source: U.S. FDA AI/ML-enabled device list source_url: https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-and-machine-learning-aiml-enabled-medical-devices source_date: 2024-05-01 tools: Lunit INSIGHT CXR; Annalise Enterprise CXR; qXR (Qure.ai) tags: worklist, computer-vision, regulation url: https://madicalaiuse.ai/en/cases/chest-xray-critical-triage url_zh: https://madicalaiuse.ai/zh-CN/cases/chest-xray-critical-triage Several cleared products flag possible urgent findings on a chest radiograph, such as pneumothorax, and move that study up a reading worklist. Names often cited for this job include Lunit INSIGHT CXR, Annalise Enterprise CXR, and qXR. This card describes the shared workflow, not a head-to-head, and it does not pool accuracy numbers across vendors. A flagged study still needs a radiologist’s report. Setting: Emergency departments, inpatient wards, and outpatient imaging lists where a portable or departmental chest radiograph is acquired before a radiologist’s formal read. Reported: No pooled performance figure. Each product’s labeled use and published accuracy differ; check the specific clearance rather than a category average. ZH-CN title: 胸部 X 线危急征象的工作列表分诊 多种已获许可的产品会标出胸片上可能的紧急所见（例如气胸），并把该检查在读片列表中提前。常被提到的包括 Lunit INSIGHT CXR、Annalise Enterprise CXR 和 qXR。本条描述的是共同工作流，不是头对头比较，也不把各厂商的准确率混成一个数字。被标记的检查仍需放射科医师出具报告。 ---- # TREWS sepsis early warning on the inpatient ward slug: trews-sepsis-early-warning category: clinical-decision-support evidence: published source_type: paper year: 2022 source: Adams et al., Nature Medicine source_url: https://www.nature.com/articles/s41591-022-01894-0 source_date: 2022-07-21 tools: TREWS; Bayesian Health tags: sepsis, early-warning, human-in-the-loop url: https://madicalaiuse.ai/en/cases/trews-sepsis-early-warning url_zh: https://madicalaiuse.ai/zh-CN/cases/trews-sepsis-early-warning The Targeted Real-time Early Warning System (TREWS) watches the electronic record and asks a clinician to confirm or dismiss a possible sepsis alert. It was studied prospectively across five hospitals, not as a silent background score. The published association depends on a person acting on the alert within a few hours. It does not diagnose sepsis by itself. Setting: Emergency departments and inpatient wards in two academic and three community hospitals, inside the existing record and antibiotic workflow. Reported: Among patients with sepsis identified before antibiotics, confirmation of the alert within 3 hours was associated with an adjusted absolute mortality difference of 3.3 percentage points (relative 18.7%) versus alerts not confirmed in that window (Nature Medicine, 2022). A companion paper reported a 1.85-hour shorter median time to the first antibiotic order for the same comparison. These are associations with how the alert was used, not a randomized trial of the model. ZH-CN title: 住院场景中的 TREWS 脓毒症早期预警 靶向实时早期预警系统（TREWS）监视电子病历，并请临床医生确认或忽略一条可能的脓毒症警报。它是在五家医院前瞻性部署后研究的，不是一个无人查看的后台分数。已发表的关联取决于有人在数小时内处理警报。系统本身不下脓毒症诊断。 ---- # External validation of a widely deployed sepsis model slug: epic-sepsis-model-validation category: clinical-decision-support evidence: published source_type: paper year: 2021 source: Wong et al., JAMA Internal Medicine source_url: https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2781307 source_date: 2021-06-21 tools: Epic Sepsis Model tags: sepsis, cautionary, early-warning url: https://madicalaiuse.ai/en/cases/epic-sepsis-model-validation url_zh: https://madicalaiuse.ai/zh-CN/cases/epic-sepsis-model-validation The Epic Sepsis Model is a proprietary score switched on in a large number of U.S. hospitals. Researchers at Michigan Medicine compared it with sepsis in their own hospitalizations, instead of relying on the vendor performance sheet. The paper is included here because a familiar model can still fail local validation. Widespread deployment is not the same thing as demonstrated usefulness. Setting: Adult hospitalizations at Michigan Medicine, with the score recalculated through the stay and compared with sepsis defined from surveillance criteria and the record. Reported: Across 38,455 hospitalizations, the hospitalization-level AUC was 0.63 (95% CI 0.62–0.64), below the discrimination the developer had reported. At a score of 6 or higher the model alerted on 18% of hospitalizations, did not identify 67% of sepsis cases, and found only 7% of sepsis patients who had not already received timely antibiotics. ZH-CN title: 对一套广泛部署的脓毒症模型做外部验证 Epic 脓毒症模型是一套专有评分，已在大量美国医院打开。密歇根医学中心的研究者用本院住院数据核对其是否真能预测脓毒症，而不是沿用厂商性能说明。把这篇论文收进目录，是因为熟悉的模型仍可能通不过本地验证。部署范围广，不等于已经被证明有用。 ---- # Suppressing non-actionable alarms in the ICU slug: icu-false-alarm-suppression category: icu-critical-care evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-09-01 tools: Multiparameter monitor; Local rules engine tags: alarms, human-in-the-loop, safety-boundary url: https://madicalaiuse.ai/en/cases/icu-false-alarm-suppression url_zh: https://madicalaiuse.ai/zh-CN/cases/icu-false-alarm-suppression This is an illustrative composite, not a trial report. A unit feeds monitor alarms through a second layer that withholds repeats already explained by motion, a loose lead, or a value the nurse just acknowledged. The bedside alarm for a new life-threatening rhythm still sounds. The point of the scenario is the governance question: who is allowed to silence what, and how a missed true alarm would be reviewed. Setting: A mixed medical-surgical ICU where nurses already carry alarm fatigue, and any suppression rule is signed off by the unit medical director and biomedical engineering. Reported: none (illustrative or not stated) ZH-CN title: 减少重症监护中无需处理的报警 这是示意性组合情景，不是试验报告。病房把监护仪报警再过一层：对体动、导联松脱，或护士刚刚确认过的数值，不再重复鸣响。新出现的致命性心律仍然立刻报警。这个情景要问的是治理问题：谁有权关闭哪一类声音，以及一旦漏掉真报警将如何复盘。 ---- # A deterioration score for the rapid-response huddle slug: ward-deterioration-huddle category: icu-critical-care evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-05-16 tools: EHR deterioration score; NEWS2 tags: early-warning, nursing, human-in-the-loop url: https://madicalaiuse.ai/en/cases/ward-deterioration-huddle url_zh: https://madicalaiuse.ai/zh-CN/cases/ward-deterioration-huddle This is an illustrative composite, not a trial report. Each morning the rapid-response nurse opens a list that combines a NEWS2-style vital-sign score with a model score from the record. The huddle decides who needs a bedside review. The score does not page a code team, order labs, or write a note. Hospitals using a vendor deterioration index should repeat a local check before trusting the threshold, as the sepsis literature already shows. Setting: Adult wards that already calculate an early-warning score, with a rapid-response nurse who can walk to the bedside the same shift. Reported: none (illustrative or not stated) ZH-CN title: 供快速反应晨会使用的病情恶化评分 这是示意性组合情景，不是试验报告。每天早晨，快速反应护士打开一份名单，把类似 NEWS2 的生命体征评分和病历模型分放在一起。晨会决定谁需要床旁复核。该分数不会自行呼叫抢救、开检验或写病程。医院若使用厂商的恶化指数，应先做本地核对再相信阈值；脓毒症文献已经说明为何要这样做。 ---- # Ambient documentation during the outpatient visit slug: ambient-clinical-documentation category: documentation-coding evidence: vendor-reported source_type: vendor year: 2025 source: Abridge source_url: https://www.abridge.com/ source_date: 2025-03-01 tools: Abridge; Nuance DAX Copilot; Nabla tags: documentation, language-model, human-in-the-loop url: https://madicalaiuse.ai/en/cases/ambient-clinical-documentation url_zh: https://madicalaiuse.ai/zh-CN/cases/ambient-clinical-documentation Ambient systems record a consented visit and draft a note the clinician edits before signing. Abridge, Nuance DAX Copilot, and Nabla are three products sold for this job. Vendors describe less after-hours charting. Independent results differ by specialty, and a draft is not a signed note. Patients need to know the visit is being captured. Setting: Outpatient clinics, often primary care and selected specialties, where the clinician reviews the draft in the record before the note is final. Reported: No single time-saved figure is repeated here. Vendor studies and early health-system reports are not interchangeable, and this index does not treat them as one result. ZH-CN title: 门诊就诊中的环境音临床文书 环境音系统在获得同意后录下就诊对话，生成临床医生签署前需要修改的病历草稿。Abridge、Nuance DAX Copilot 和 Nabla 是面向这一用途销售的三款产品。厂商描述下班后写病历的时间减少。独立结果因专科而异，草稿也不是已签名病历。患者需要知道本次就诊正在被采集。 ---- # Computer-assisted coding with a coder still accountable slug: coder-in-the-loop-coding category: documentation-coding evidence: vendor-reported source_type: vendor year: 2024 source: Solventum source_url: https://www.solventum.com/ source_date: 2024-01-15 tools: Solventum computer-assisted coding; 3M CAC heritage tags: coding, admin, human-in-the-loop url: https://madicalaiuse.ai/en/cases/coder-in-the-loop-coding url_zh: https://madicalaiuse.ai/zh-CN/cases/coder-in-the-loop-coding Computer-assisted coding suggests diagnosis and procedure codes from the note and sends them to a professional coder. Products in the 3M lineage, now under Solventum, have done a version of this for years, and newer language models are being added on top. The coder remains the person who submits the claim. A suggestion is not documentation of a condition the clinician did not record. Setting: Hospital health-information management and physician-group coding teams, after the clinical note exists and before the claim goes out. Reported: No revenue or accuracy figure is stated. Productivity claims from vendors should be read as vendor-reported. ZH-CN title: 编码员仍负最终责任的计算机辅助编码 计算机辅助编码从病历中提出诊断和操作编码建议，交给专业编码员。源自 3M、现属 Solventum 的产品已做这类工作多年，新的语言模型正在叠加上去。提交账单的仍然是编码员。建议不能用来补记临床医生没有写过的诊断。 ---- # Parallel notification of a suspected large-vessel occlusion slug: viz-contact-lvo category: triage-emergency evidence: regulatory source_type: news year: 2018 source: U.S. FDA announcement, 13 February 2018 source_url: https://www.fda.gov/news-events/press-announcements/fda-permits-marketing-clinical-decision-support-software-alerting-providers-potential-stroke source_date: 2018-02-13 tools: Viz.ai ContaCT tags: stroke, emergency, regulation, computer-vision url: https://madicalaiuse.ai/en/cases/viz-contact-lvo url_zh: https://madicalaiuse.ai/zh-CN/cases/viz-contact-lvo Viz.ai ContaCT analyzes brain CT angiography and can text a neurovascular specialist when it suspects a large-vessel occlusion, in parallel with the usual read. FDA permitted marketing in 2018 as notification software. The specialist still reviews full images on a diagnostic workstation. The product is not a diagnosis and not a treatment decision. Setting: Emergency stroke imaging, for adults who already have a CT angiogram as part of a stroke workup, inside a hospital network that has a specialist on call. Reported: FDA described a retrospective set of 300 CT angiograms reviewed against neuro-radiologists, plus real-world evidence aimed at earlier specialist notification. This card does not restate a sensitivity figure. The labeled limit is notification, not a stand-alone read. ZH-CN title: 可疑大血管闭塞的并行通知 Viz.ai ContaCT 分析头颅 CT 血管成像，若怀疑大血管闭塞，可在常规读片的同时向神经血管专科医生发送通知。FDA 于 2018 年允许其作为通知类软件上市。专科医生仍须在诊断工作站上查看完整图像。该产品不是诊断，也不是治疗决定。 ---- # A suggested emergency acuity for the triage nurse slug: ed-triage-acuity-suggestion category: triage-emergency evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-08-20 tools: EHR triage form; Local acuity model tags: emergency, human-in-the-loop, safety-boundary url: https://madicalaiuse.ai/en/cases/ed-triage-acuity-suggestion url_zh: https://madicalaiuse.ai/zh-CN/cases/ed-triage-acuity-suggestion This is an illustrative composite, not a trial report. As the triage nurse types the chief complaint and first vitals, a model suggests an acuity level. The nurse accepts or overrides it, and the accepted level is the one that enters the queue. The scenario is about keeping a human on the decision when under-triage would delay care. No accuracy percentage is claimed, because this entry is not a study. Setting: A hospital emergency department using a five-level acuity scale, with the suggestion visible only to the triage nurse. Reported: none (illustrative or not stated) ZH-CN title: 供分诊护士参考的急诊分诊级别建议 这是示意性组合情景，不是试验报告。分诊护士录入主诉和首次生命体征时，模型给出一个分诊级别建议。护士接受或改写，进入队列的是护士确认后的级别。这个情景关心的是：分诊过低会耽误救治，因此决定必须留在人手里。此处不声称任何准确率，因为本条不是一项研究。 ---- # Adjunct detection on prostate biopsy whole-slide images slug: paige-prostate-assist category: pathology evidence: regulatory source_type: news year: 2021 source: U.S. FDA De Novo DEN200080 source_url: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/denovo.cfm?id=DEN200080 source_date: 2021-09-21 tools: Paige Prostate; Paige FullFocus tags: pathology, regulation, computer-vision, human-in-the-loop url: https://madicalaiuse.ai/en/cases/paige-prostate-assist url_zh: https://madicalaiuse.ai/zh-CN/cases/paige-prostate-assist Paige Prostate marks a location on a digitized prostate biopsy slide that the model ranks as the most suspicious for cancer, after the pathologist’s own first look. FDA authorized it in 2021 as an adjunct. The output is not the diagnosis. The pathologist still signs out, using the full clinical context and any extra stains they choose to order. Setting: Surgical pathology laboratories reading hematoxylin-and-eosin prostate needle biopsies that have been scanned on a supported whole-slide scanner. Reported: FDA reported that, across 16 pathologists and 527 slide images, cancer detection on individual slides improved by 7.3% on average versus unassisted reads, with no effect on the read of benign slides. The study did not measure the effect on the final patient diagnosis, which usually rests on more than one core. ZH-CN title: 前列腺穿刺全切片上的辅助检出 Paige Prostate 在病理医生完成第一遍阅片后，于数字化前列腺穿刺切片上标出一个模型认为最可疑为癌的位置。FDA 于 2021 年将其作为辅助工具授权。输出不是诊断。病理医生仍结合全部临床信息和其决定加做的染色来签发报告。 ---- # Quantitative assist for immunohistochemistry scoring slug: ihc-quantification-assist category: pathology evidence: vendor-reported source_type: vendor year: 2025 source: PathAI source_url: https://www.pathai.com/ source_date: 2025-02-01 tools: PathAI; Ibex tags: pathology, human-in-the-loop, computer-vision url: https://madicalaiuse.ai/en/cases/ihc-quantification-assist url_zh: https://madicalaiuse.ai/zh-CN/cases/ihc-quantification-assist Vendors including PathAI and Ibex offer algorithms that estimate staining extent, such as a Ki-67 or PD-L1 percentage, on a digitized slide. A pathologist checks the region and the number before it enters the report. This card does not restate concordance percentages, which are assay-specific and easy to over-read. The signed score remains a pathology result, not a model result. Setting: Immunohistochemistry sign-out in a digital pathology lab, where the stain already has a clinical scoring scheme the tumor board expects. Reported: No concordance figure is copied here. Treat vendor white papers as vendor-reported until an independent study matches the assay you actually run. ZH-CN title: 免疫组化评分的定量辅助 包括 PathAI 和 Ibex 在内的厂商提供算法，用于估计数字化切片上的染色范围，例如 Ki-67 或 PD-L1 百分比。病理医生核对区域和数字后，结果才进入报告。本条不转述一致性百分比，那些数字依赖具体检测，且很容易被读过头。签发的评分仍是病理结果，不是模型结果。 ---- # A generative-AI-designed molecule in a phase 2a IPF trial slug: rentosertib-ipf category: drug-discovery-pharma evidence: published source_type: paper year: 2025 source: Xu et al., Nature Medicine source_url: https://doi.org/10.1038/s41591-025-03743-2 source_date: 2025-06-03 tools: Pharma.AI; Insilico Medicine; rentosertib (ISM001-055) tags: drug-discovery, trials url: https://madicalaiuse.ai/en/cases/rentosertib-ipf url_zh: https://madicalaiuse.ai/zh-CN/cases/rentosertib-ipf Insilico Medicine used a generative platform to propose TNIK as a target in idiopathic pulmonary fibrosis and to design rentosertib (formerly ISM001-055). Nature Medicine published a 12-week randomized phase 2a trial in 2025. The primary endpoint was safety, not lung function. The molecule is investigational. A small secondary signal is not an approved therapy. Setting: A multicenter, double-blind, placebo-controlled trial in adults with idiopathic pulmonary fibrosis, run at sites in China (NCT05938920). Reported: Seventy-one patients were randomized. Treatment-emergent adverse events were in a similar range across arms (about 71–83%). The most discussed discontinuation events involved liver-test abnormalities or diarrhea. At 60 mg once daily, mean forced vital capacity changed by +98.4 mL versus −20.3 mL on placebo. That lung-function comparison was secondary, the samples were small, and the authors said it needs a larger, longer trial. ZH-CN title: 生成式人工智能设计的分子进入特发性肺纤维化 2a 期试验 Insilico Medicine 用生成式平台提出 TNIK 作为特发性肺纤维化的靶点，并设计了 rentosertib（原名 ISM001-055）。《自然·医学》于 2025 年发表了一项 12 周随机 2a 期试验。主要终点是安全性，不是肺功能。该分子仍在研究中。一个很小的次要信号不是已批准疗法。 ---- # Matching a cancer chart to trial eligibility slug: oncology-trial-matching category: drug-discovery-pharma evidence: vendor-reported source_type: vendor year: 2025 source: Tempus source_url: https://www.tempus.com/ source_date: 2025-04-01 tools: Tempus; Flatiron; Deep 6 AI tags: trials, human-in-the-loop, language-model url: https://madicalaiuse.ai/en/cases/oncology-trial-matching url_zh: https://madicalaiuse.ai/zh-CN/cases/oncology-trial-matching Trial-matching tools read structured fields and sometimes the free-text note to list studies a patient might qualify for. Tempus, Flatiron, and Deep 6 AI are examples of companies that sell a version of this. A research coordinator still checks every criterion against the chart. A match list is not enrollment, and it is not a treatment recommendation. Setting: Oncology clinics and trial offices that already have a coordinator, using the match list to decide whom to screen by hand. Reported: No enrollment-rate or speed figure is repeated here. Those claims are vendor-specific and often omit how many suggestions a coordinator rejects. ZH-CN title: 把肿瘤病历与试验入组标准相匹配 试验匹配工具读取结构化字段，有时也读自由文本病程，列出患者可能符合的研究。Tempus、Flatiron 和 Deep 6 AI 是销售此类能力的例子。研究协调员仍须对照病历逐条核对。匹配列表不是入组，也不是治疗建议。 ---- # A home signal that prompts a heart-failure nurse to call slug: heart-failure-nurse-flag category: remote-monitoring evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-04-22 tools: Home weight and blood pressure; Clinic message pool tags: remote-care, nursing, early-warning url: https://madicalaiuse.ai/en/cases/heart-failure-nurse-flag url_zh: https://madicalaiuse.ai/zh-CN/cases/heart-failure-nurse-flag This is an illustrative composite, not a trial report. Daily weight, blood pressure, and a short symptom check land in a pool. A simple model ranks who should be called today when the heart-failure nurse starts the shift. The nurse places the call and decides on a diuretic plan with the covering clinician. The scenario deliberately stops before any claim that admissions fell. Device trials of hemodynamic monitors are a different evidence base and are not borrowed here. Setting: A heart-failure clinic that already enrolls selected patients in home monitoring and staffs a nurse to review the queue on weekdays. Reported: none (illustrative or not stated) ZH-CN title: 提示心衰护士主动来电的居家信号 这是示意性组合情景，不是试验报告。每日体重、血压和简短症状核对进入一个队列。心衰护士开始当班时，一个简单模型排列今天该先给谁打电话。电话由护士打出，利尿方案与值班医生一起决定。这个情景刻意停在“住院是否减少”之前。血流动力学植入装置的试验是另一类证据，这里不借用。 ---- # A weekly glucose-pattern note for the diabetes visit slug: cgm-clinic-summary category: remote-monitoring evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-09-14 tools: Continuous glucose monitor export; Clinic inbox tags: remote-care, safety-boundary, documentation url: https://madicalaiuse.ai/en/cases/cgm-clinic-summary url_zh: https://madicalaiuse.ai/zh-CN/cases/cgm-clinic-summary This is an illustrative composite, not a trial report. Before a diabetes visit, software drafts a short note from a continuous glucose export: time in range, overnight lows, and the days that look different. The clinician edits it and decides any change in insulin. The scenario excludes autonomous dosing. Closed-loop insulin systems are regulated devices with their own labels and are not this workflow. Setting: Endocrinology or primary care diabetes clinics that already receive glucose downloads and want the pattern summarized before the conversation. Reported: none (illustrative or not stated) ZH-CN title: 供糖尿病门诊使用的一周血糖模式摘要 这是示意性组合情景，不是试验报告。糖尿病随诊前，软件根据持续葡萄糖监测导出数据起草一段短记录：目标范围内时间、夜间低血糖，以及看起来不同的那几天。临床医生修改草稿，并决定胰岛素是否调整。这个情景不包括自动给药。闭环胰岛素系统是有独立说明书的监管器械，不是这条工作流。 ---- # A plain-language discharge explanation the clinician edits slug: plain-language-discharge category: patient-engagement evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-09-22 tools: EHR discharge note; Clinician edit step tags: patient-facing, language-model, human-in-the-loop url: https://madicalaiuse.ai/en/cases/plain-language-discharge url_zh: https://madicalaiuse.ai/zh-CN/cases/plain-language-discharge This is an illustrative composite, not a trial report. At discharge, a model drafts a short explanation of the diagnosis, medicines, and warning signs from the clinician’s own note, in the patient’s language where the hospital supports it. The clinician or pharmacist edits every line before it is printed or sent. The draft is not allowed to add a new instruction that is absent from the signed note. It is not a chatbot the patient can ask for new medical advice. Setting: Inpatient discharge on a medical ward, with a clinician who is already responsible for the after-visit summary. Reported: none (illustrative or not stated) ZH-CN title: 由临床医生审改的通俗出院说明 这是示意性组合情景，不是试验报告。出院时，模型依据临床医生自己的记录，用医院支持的患者语言起草一段简短说明，涵盖诊断、药物和警示症状。临床医生或药师在打印或发送前逐句修改。草稿不得加入已签名记录中没有的新医嘱。它不是一个可以让患者继续追问新医学建议的聊天机器人。 ---- # A patient-facing symptom assessment, not a diagnosis slug: symptom-assessment-app category: patient-engagement evidence: vendor-reported source_type: vendor year: 2024 source: Ada Health source_url: https://ada.com/ source_date: 2024-01-01 tools: Ada tags: patient-facing, safety-boundary url: https://madicalaiuse.ai/en/cases/symptom-assessment-app url_zh: https://madicalaiuse.ai/zh-CN/cases/symptom-assessment-app Ada is a symptom-assessment app that interviews a person and suggests possible causes plus a level of urgency. It is a patient-facing product, not a clinician order. Regulatory status and intended use differ by country. An assessment is not a diagnosis, not a prescription, and not a reason to delay emergency care when red-flag symptoms are present. Setting: People using a phone before they decide whether to seek care, and some health systems that offer the same interview as a front door. A clinician is not in the loop unless the person is handed off. Reported: No diagnostic-accuracy figure is copied here. Published evaluations exist and disagree in method; they should be read individually rather than as a single sensitivity. ZH-CN title: 面向患者的症状评估，而不是诊断 Ada 是一款症状评估应用：向使用者提问，然后给出可能的原因和紧急程度。它面向患者，不是临床医嘱。监管状态和预期用途因国家而异。评估不是诊断，不是处方，也不能在出现危险症状时成为推迟急诊的理由。 ---- # A guided CBT companion between therapy sessions slug: guided-cbt-companion category: mental-health evidence: vendor-reported source_type: vendor year: 2024 source: Wysa source_url: https://www.wysa.com/ source_date: 2024-06-01 tools: Wysa tags: mental-health, patient-facing, safety-boundary url: https://madicalaiuse.ai/en/cases/guided-cbt-companion url_zh: https://madicalaiuse.ai/zh-CN/cases/guided-cbt-companion Wysa is an example of a conversational program that offers cognitive-behavioral exercises between visits with a person. It is support, not a therapist, and it is not a crisis service. Someone at immediate risk needs local emergency care or a crisis line, not a chatbot. Health systems that offer a tool like this still need a path to a human. Setting: Outpatient mental health or employee and student programs, used between scheduled sessions, with clear wording that the app does not handle emergencies. Reported: No depression-score change is repeated here. Trial results for digital CBT vary by product, population, and whether a human coach is included. ZH-CN title: 治疗间歇期的引导式认知行为支持 Wysa 是一例对话式程序，在与真人治疗的间歇提供认知行为练习。它是支持，不是治疗师，也不是危机服务。处于即时危险的人需要当地急救或危机热线，而不是聊天机器人。提供这类工具的医疗系统仍然需要一条找到真人的路径。 ---- # REACH VET outreach for veterans at highest statistical risk slug: reach-vet-outreach category: mental-health evidence: published source_type: paper year: 2023 source: McCarthy et al., Psychiatric Services source_url: https://psychiatryonline.org/doi/10.1176/appi.ps.202100629 source_date: 2023-02-01 tools: REACH VET; VHA electronic health record tags: mental-health, population, human-in-the-loop url: https://madicalaiuse.ai/en/cases/reach-vet-outreach url_zh: https://madicalaiuse.ai/zh-CN/cases/reach-vet-outreach REACH VET is a U.S. Veterans Health Administration program. A statistical model flags veterans in a very small top tier of risk each month, and a clinician who knows the patient reviews the chart and makes contact. Identification is not a diagnosis of suicidal intent. The program exists to prompt a human conversation and a look at whether care should change. Setting: Veterans Health Administration facilities. The model runs monthly on patients seen in the prior two years. Policy expects outreach to the top 0.1% tier at each facility. Reported: A program evaluation reported that by late 2020 nearly all identified veterans had been contacted and had care reviewed. The same paper associated the program, after age and sex adjustment, with a 5% lower probability of a documented suicide attempt, plus more outpatient contact and safety planning. That is a program association, not a statement that the score predicts an individual death. ZH-CN title: 面向统计风险最高退伍军人的 REACH VET 主动联系 REACH VET 是美国退伍军人健康管理局的项目。统计模型每月标出风险处于极小顶层的退伍军人，熟悉该患者的临床医生复核病历并主动联系。被标出不是自杀意图的诊断。项目的作用是促成一次人与人的谈话，并查看治疗是否需要调整。 ---- # A next-shift workload forecast for the charge nurse slug: shift-acuity-forecast category: nursing-workflow evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-02-02 tools: Staffing office workbook; Ward census tags: nursing, admin url: https://madicalaiuse.ai/en/cases/shift-acuity-forecast url_zh: https://madicalaiuse.ai/zh-CN/cases/shift-acuity-forecast This is an illustrative composite, not a trial report. Late in the afternoon a model estimates how heavy the next shift will be from the current census, planned admissions, and recent acuity. The charge nurse uses it as one input to ask the staffing office for help. It does not cut a shift by itself, and it does not replace the legal and contractual staffing rules of the hospital. No savings percentage is stated. Setting: Inpatient units with a charge nurse and a central staffing office that can still move people before the shift starts. Reported: none (illustrative or not stated) ZH-CN title: 供责任护士参考的下一班工作量预测 这是示意性组合情景，不是试验报告。下午晚些时候，模型根据当前在院人数、计划入院和近期危重程度估计下一班有多忙。责任护士把它作为向护理排班办公室求助的一项输入。它不能自行砍掉一个班次，也不能代替医院在法律和合同上的人员配置规则。此处不给出节省比例。 ---- # Fall-risk context on the bedside rounding list slug: fall-round-prompts category: nursing-workflow evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-08-03 tools: EHR fall-risk score; Bedside rounding list tags: nursing, early-warning, safety-boundary url: https://madicalaiuse.ai/en/cases/fall-round-prompts url_zh: https://madicalaiuse.ai/zh-CN/cases/fall-round-prompts This is an illustrative composite, not a trial report. The rounding list shows who is at higher fall risk and which precautions are already ordered, so the nurse does not hunt through the chart in the doorway. It is a prompt, not a new score the nurse must obey. Alert fatigue is part of the design problem: if every patient is flagged, the list has failed. No fall-rate reduction is claimed. Setting: Adult medical and surgical wards during scheduled bedside rounds, using a fall-risk assessment the hospital already documents. Reported: none (illustrative or not stated) ZH-CN title: 床旁巡视清单上的跌倒风险提示 这是示意性组合情景，不是试验报告。巡视清单显示谁的跌倒风险较高、哪些预防措施已经开立，护士不必在门口翻病历。它是提示，不是护士必须服从的新分数。报警疲劳是设计问题的一部分：如果每个人都被标记，清单就失败了。此处不声称跌倒率下降。 ---- # A prior-authorization packet drafted from the chart slug: prior-auth-packet-draft category: revenue-cycle evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-10-12 tools: EHR chart; Payer portal tags: admin, language-model, human-in-the-loop url: https://madicalaiuse.ai/en/cases/prior-auth-packet-draft url_zh: https://madicalaiuse.ai/zh-CN/cases/prior-auth-packet-draft This is an illustrative composite, not a trial report. For a planned imaging study or procedure, a model assembles the indication, prior treatments, and relevant results into a draft the authorization staff edit before anything is sent to a payer. Staff have to check that each sentence is actually in the chart. Models omit and sometimes invent. No approval-rate improvement is claimed. Setting: A clinic or hospital authorization team working ahead of elective studies, with the ordering clinician available when the clinical story is unclear. Reported: none (illustrative or not stated) ZH-CN title: 依据病历草拟的事前授权材料 这是示意性组合情景，不是试验报告。对一项计划中的影像或操作，模型把适应证、既往治疗和相关结果整理成草稿，授权人员在发给支付方之前修改。每一句话是否真的写在病历里，必须由人工核对。模型会遗漏，有时也会编造。此处不声称批准率提高。 ---- # An appeal letter drafted from the denial and the note slug: denial-appeal-draft category: revenue-cycle evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-07-27 tools: Denial letter; Clinical note tags: admin, language-model, coding url: https://madicalaiuse.ai/en/cases/denial-appeal-draft url_zh: https://madicalaiuse.ai/zh-CN/cases/denial-appeal-draft This is an illustrative composite, not a trial report. After a payer denies a claim, a model drafts an appeal that quotes the denial reason and the passages in the note that speak to it. A coder or clinician edits the letter and decides whether the appeal is worth sending. The draft cannot create a diagnosis that was never documented. No overturn rate is claimed. Setting: Hospital revenue-cycle or clinic billing teams, after a written denial, with access to the original note and the coding summary. Reported: none (illustrative or not stated) ZH-CN title: 依据拒付函和病历草拟的申诉信 这是示意性组合情景，不是试验报告。支付方拒付后，模型草拟一封申诉信，引用拒付理由以及病历中对应的段落。编码员或临床医生修改信件，并决定是否值得发出。草稿不能制造一份从未记录过的诊断。此处不声称翻盘率。 ---- # Recognizing surgical phases in laparoscopic video slug: endonet-phase-recognition category: surgery-robotics evidence: published source_type: paper year: 2016 source: Twinanda et al., arXiv:1602.03012 source_url: https://arxiv.org/abs/1602.03012 source_date: 2016-02-09 tools: EndoNet; Cholec80 tags: surgery, research, computer-vision url: https://madicalaiuse.ai/en/cases/endonet-phase-recognition url_zh: https://madicalaiuse.ai/zh-CN/cases/endonet-phase-recognition EndoNet, from Twinanda and colleagues, is a research network that recognizes surgical phases and instruments in laparoscopic cholecystectomy video. It was demonstrated on the public Cholec80 collection. The result is about whether a model can tell where a case is in its sequence. It is not evidence that video analysis reduces complications, and it is not a product cleared to guide a surgeon’s hands. Setting: Recorded laparoscopic cholecystectomy video used for computer-vision research, not a live advisory in the operating room. Reported: The paper reports phase and tool recognition performance on Cholec80. Those research metrics are not restated here as clinical outcomes. Read the paper for the figures. ZH-CN title: 在腹腔镜视频中识别手术阶段 Twinanda 等人的 EndoNet 是一个研究用网络，用于在腹腔镜胆囊切除视频中识别手术阶段和器械。它在公开的 Cholec80 数据集上做了演示。结果回答的是：模型能否判断手术进行到哪一段。它不是“视频分析能减少并发症”的证据，也不是获准引导外科医生双手的产品。 ---- # Post-case analytics for robotic surgery coaching slug: robotic-case-coaching category: surgery-robotics evidence: vendor-reported source_type: vendor year: 2025 source: Intuitive source_url: https://www.intuitive.com/ source_date: 2025-01-15 tools: da Vinci robotic platform; Post-case reports tags: surgery, human-in-the-loop url: https://madicalaiuse.ai/en/cases/robotic-case-coaching url_zh: https://madicalaiuse.ai/zh-CN/cases/robotic-case-coaching Robotic systems such as the da Vinci platform can store console time, instrument use, and in some setups video that a surgeon reviews after the case. Vendors present this as coaching and quality review. The surgeon still operates. Nothing in this card says that analytics lower complications. That would be a separate study of a defined program. Setting: Operating rooms already using a robotic system, with review happening after the patient is out of the room, often in teaching conferences. Reported: No complication or learning-curve figure is stated. Vendor dashboards describe activity, not patient outcomes, unless a specific study says otherwise. ZH-CN title: 用于机器人手术带教的术后病例分析 达芬奇等机器人系统可以保存控制台时间、器械使用情况，在部分配置下还有供外科医生术后回看的视频。厂商将其表述为带教和质量复核。手术仍由外科医生完成。本条没有说分析能降低并发症。那需要针对一个明确项目的单独研究。 ---- # A wastewater trend brief for the public health desk slug: wastewater-trend-brief category: public-health evidence: program source_type: news year: 2024 source: U.S. CDC National Wastewater Surveillance System source_url: https://www.cdc.gov/nwss/ source_date: 2024-09-01 tools: CDC NWSS; Local epidemiology desk tags: population, early-warning url: https://madicalaiuse.ai/en/cases/wastewater-trend-brief url_zh: https://madicalaiuse.ai/zh-CN/cases/wastewater-trend-brief The U.S. CDC National Wastewater Surveillance System publishes pathogen trends from sewage, which health departments read next to emergency and lab data. Some jurisdictions add a statistical nowcast on top of that feed. This card is about the desk workflow, not a claim that a named model forecasts incidence with a stated error. Wastewater is a population signal. It does not diagnose a person. Setting: A city or state public health epidemiology desk during respiratory-virus season, briefing hospitals on whether trends are rising. Reported: No model error rate is quoted. Use the CDC program pages and the local health department method note for what a given jurisdiction actually publishes. ZH-CN title: 供公共卫生值班使用的污水监测趋势简报 美国 CDC 国家污水监测系统发布污水中的病原体趋势，卫生部门将其与急诊和实验室数据对照阅读。部分地区会在这条数据上再加统计现时预测（nowcast）。本条讲的是值班工作流，不是声称某个具名模型以某一误差预测发病率。污水是人群信号，不能诊断个人。 ---- # A navigator list for people overdue for cancer screening slug: screening-gap-outreach category: public-health evidence: illustrative source_type: hospital year: 2026 source: Illustrative composite, Madical AI Use source_url: https://madicalaiuse.ai/en/about#evidence source_date: 2026-03-18 tools: Population health registry; Outreach workqueue tags: population, equity, screening url: https://madicalaiuse.ai/en/cases/screening-gap-outreach url_zh: https://madicalaiuse.ai/zh-CN/cases/screening-gap-outreach This is an illustrative composite, not a trial report. A registry ranks patients who are overdue for a fit test, mammogram, or other screening the health system already recommends. Navigators call down the list. Before anyone trusts the rank order, the team checks whether the score under-calls a language group, a neighborhood, or a payer. No detection-rate improvement is claimed. A biased rank can widen a gap the program meant to close. Setting: A health-system population health team with navigators, working from an existing screening registry rather than from a new disease label. Reported: none (illustrative or not stated) ZH-CN title: 为逾期癌症筛查生成访视员名单 这是示意性组合情景，不是试验报告。登记系统对已逾期、且医疗系统本就建议做粪便潜血、钼靶或其他筛查的患者排序。访视员按名单打电话。在信任排序之前，团队要检查分数是否少叫到某一语言群体、街区或支付类型。此处不声称检出率提高。有偏的排序会加大项目本来想缩小的差距。 ----