起步顺序Order of work 先理清隐私与用途,再与一家医院试点,之后做公共卫生;检验流程 AI 同步推进。Settle privacy and purpose first, pilot with one hospital, then public health, with lab workflow AI alongside.
公开事实Public facts 已核对Checked
- 380+三省采集中心数量Collection centres across three provincesBC、安省、萨省;另有 13 家实验室BC, Ontario and Saskatchewan; 13 labs
- 100M+每年检验次数Lab tests a year公司口径;其他资料略有出入Company figure; other sources differ
- 16Klinrisk 肾病风险模型使用的检验项目数Lab results used by the Klinrisk kidney-risk model常规血液与尿液检测;仅安省提供Routine blood and urine tests; Ontario only
- 102AI 在历史图谱中发现的此前漏检疑似病例Missed suspected cases found by A I in old traces来自 150,000 份 SPEP 图谱;CSCC 2022 论文Among 150,000 SPEP traces; CSCC 2022 paper
需要完成的三个转变Three shifts to make 判断Judgment
- 服务药企的洞察Insight for pharma → 隐私审查之后,服务医院Insight for hospitals, after privacy review
- 逐项检测的风险评分A risk score, one test at a time → 按患者群体核对结果Results checked by patient group
- 检验 AI 停留在研究项目Lab AI as a research project → 检验 AI 进入日常流程Lab AI in daily workflow
专家组裁决Expert panel 依据公开观点的推演Inference from public views
Ziad Obermeyer
每个评分:预测什么、临床需要什么、按人群看结果。For each score: what it predicts, what care needs, results by group.
Thomas Davenport
新增医院客户会改变业务模式;点名负责人。New hospital clients change the business model; name the owner.
Andrew Ng
一家医院伙伴、一个问题,六个月出数字。One hospital partner, one question, numbers in six months.
Erik Brynjolfsson
声称 AI 省时之前,先设检验周转时间基线。Baseline lab turnaround before claiming that A I saves time.
Ethan Mollick
让检验人员测出分流与质控预警的失败点。Let technologists test where triage and QC alerts fail.
Foteini Agrafioti
数据、产品、实验室与合规的权责写成一页纸。One page of ownership: data, product, lab and compliance.
Andrew Ng:现在就与一家医院试点。Pilot with one hospital now.
Ziad Obermeyer:先理清用途与隐私。Settle purpose and privacy first.
→ 隐私审查通过后再开始试点。Pilot starts after a privacy review.
六位专家是谁、怎样思考、有什么铁律 →Who the six experts are, how they think, their rules →
三个目标Three priorities 建议Recommendation
理清隐私与用途Settle privacy and purpose
写明每个数据产品的允许用途;与隐私顾问确认。Write allowed uses for each data product; confirm with privacy counsel.
与一家医院试点Pilot with one hospital
一个问题(如疾病趋势);设基线并有明确的临床负责人。One question, such as disease trends; a baseline and a named clinical owner.
让 AI 进入检验流程Put A I into lab workflow
样本分流与质控预警,由检验人员确认。Sample triage and quality alerts, with technologists confirming.
场景组合Scenarios 建议Recommendation
| 场景Scenario | 做法Approach | 状态Status |
|---|---|---|
| 面向医院的 Health InsightsHealth Insights for hospitals 增收Revenue体验Experience | 疾病趋势预警与筛查效果评估,先过同意与隐私审查。Disease-trend alerts and screening results, after consent and privacy review. | 建议 · 基于 Health InsightsProposed · builds on Health Insights |
| 检验流程 AILab workflow A I 提效Efficiency降风险Risk | 样本分流、质控异常检测与危急值预警。Sample triage, quality anomaly detection and critical-value alerts. | 建议Proposed |
| 患者检验结果解读Result explainer for patients 体验Experience降风险Risk | 通俗语言解读与下一步建议;与临床治理和法律顾问确认。Plain-language results and next steps; confirm with clinical governance and counsel. | 建议 · 完整审查Proposed · full review |
| 智能预约与采集Smarter scheduling 提效Efficiency体验Experience | 在 380 多个采集中心优化预约与排班,减少等待。Schedule appointments and collection sites to cut waiting at 380-plus centres. | 建议Proposed |
12 个月路线12-month roadmap 建议Recommendation
奠基Foundation
- 允许用途登记册Allowed-use register
- 模型清单与分级Model inventory and tiers
- 检验周转基线Lab turnaround baseline
试点Pilot
- 一家医院伙伴One hospital partner
- 一个实验室分流预警Triage alert in one lab
- 检验人员测边界Technologists test edges
上线Go live
- 医院洞察上线Hospital insight live
- 更多站点的实验室预警Lab alerts in more sites
- 上线即监控Monitoring from day one
扩展Extend
- 公共卫生立项Scope public health
- 复核患者端工具Review patient tools
- 重排下年优先级Re-rank next year
治理与合规Governance 建议Recommendation
- 健康数据用途:与隐私顾问确认Health data uses: confirm with privacy counsel
- 患者端工具:临床治理审查Patient-facing tools: clinical governance review
- 改变检验结果的预警由检验人员确认A technologist confirms any alert that changes a result
三层指标Three-layer metrics 建议Recommendation
| 采用Adoption | 每周使用预警的检验人员数Technologists using alerts weekly · 活跃的医院合作方数Hospital partners active |
|---|---|
| 过程Process | 预警的精确率与召回率Alert precision and recall · 检验周转时间Lab turnaround time · 检验人员改判率Technologist override rate |
| 结果Outcome | 周转时间对照基线Turnaround against baseline · 评分结果按患者群体比较Score results by patient group |
目标值由企业自定。Targets are the company's to set.
风险Risks 判断Judgment
评分预测的不是想要的东西A score that predicts the wrong thing
模型可能预测谁会被检测,而非谁生病;两者都写明。A model may predict who gets tested, not who is sick; name both.
超出同意范围的数据使用Data use beyond consent
新客户意味着新用途;逐项与隐私顾问确认。New buyers mean new purposes; confirm each with privacy counsel.
同名机构混淆Confusing similar names
有无关公司同名;证据只用于本公司。Unrelated firms share the name; keep evidence about this company only.
前 90 天First 90 days 建议Recommendation
| 第 1–2 周Weeks 1–2 | 为数据、产品、实验室点名负责人Name owners for data, product, lab |
|---|---|
| 第 3–8 周Weeks 3–8 | 写明允许用途;盘点模型Write allowed uses; inventory models |
| 第 9–12 周Weeks 9–12 | 选一家医院伙伴和一个实验室Pick a hospital partner and one lab |
| 第 13 周Week 13 | 首次复盘First review |
我们可能错在:We could be wrong if: 医院已有所需数据,或隐私审查显示这些用途不被允许。hospitals already have the data they need, or privacy review shows the uses are not allowed.
本页是基于公开资料的分析,与所述企业没有合作关系,不代表其官方立场或内部计划;企业数字多来自其自述或媒体转述;「专家组」是依据六位专家公开观点的推演,不是他们的原话,也不代表他们或其雇主的立场。资料整理日期:2026-11-14。This page is an analysis based on public information. It has no affiliation with the company and does not represent its official position or internal plans. Company figures come mostly from its own statements or media reports. The expert panel is an inference from six experts' public views, not their words, and does not represent them or their employers. Compiled 2026-11-14.
想给你的公司也做一份?Want one for your company?
同样的方法:公开事实、专属判断、12 个月路线和前 90 天。公司大小都可以聊,中文沟通。Same method: public facts, a company-specific judgment, a 12-month roadmap and the first 90 days. Any size of business; we can talk in Chinese or English.