Case Study: Situation Analysis

From: Marie Jensen <m.jensen@merck.com> To: Alex Jensen <a.jensen@merck.com> Subject: Fwd: URGENT - UroScan Situation Date: Today 15:30 Priority: High Alex - Emergency surgery. Hospital network threatening media tomorrow AM about "biased AI recommendations." Crucial info about validation process in chain + attachment. Need response plan by EOD. Phone off now. –Marie ----------------------------------------------------------- Begin Forwarded Email Chain ----------------------------------------------------------- From: Women's Health Network <advocacy@whnetwork.org> To: Marie Jensen <m.jensen@merck.com> Cc: Legal Department <legal@merck.com> Subject: FINAL NOTICE - Media Release Tomorrow Date: Today 09:15 Dr. Jensen, We write to inform you that the Women’s Health Network (WHN) has decided to publish our findings on Sotelab Pharmaceuticals’s UroScan AI diagnostic system at 9:00 AM EST tomorrow. As you are aware from our prior correspondence, these findings point to a substantial performance gap between male and female patients, with approximately a 65% accuracy rate observed among female patients versus an overall (and publicly stated) ~95% figure in other demographics. We consider this a critical public health issue. Our mission is to ensure that medical technologies are held to the highest standards of equity and accuracy. UroScan’s reported discrepancy raises grave concerns about patient safety and fairness in healthcare delivery. Should Sotelab Pharmaceuticals wish to provide an official statement or clarifications for inclusion in our release, we require it no later than midnight tonight. We sincerely hope you will address these concerns transparently. Sincerely, Dr. Sarah Martinez Director of Medical Ethics Women’s Health Network --- From: James Wilson <j.wilson@legal.merck.com> To: Marie Jensen <m.jensen@merck.com> Subject: Re: UroScan Performance Concerns Date: Yesterday 16:45 Marie, I’ve reviewed the initial internal notes regarding the performance discrepancy—specifically, the serious allegations about UroScan’s lower accuracy for female patients (65% vs. an expected 95%). This situation poses major legal and reputational risks for Sotelab Pharmaceuticals if not handled swiftly and thoughtfully. As Chief Legal Officer, I advise the following: 1. Centralize all communications through Legal to ensure consistency. 2. Verify the underlying data that the Women’s Health Network references. 3. Prepare a cautious statement highlighting our commitment to patient safety and ongoing investigations. We cannot afford any perception that we are dismissing these claims. Please coordinate closely with my office on all external messaging. I’ll await further updates from your team or from Medical Affairs leadership. Regards, James Wilson Chief Legal Officer, Sotelab Pharmaceuticals Medical Solutions --- From: Central Hospital Data Committee <data@central-hospital.org> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: UroScan Performance Analysis Date: 5 days ago 14:20 Dear Implementation Team, After gathering three weeks of UroScan performance data at Central Hospital, our committee has identified a notable discrepancy across different patient groups. Namely, we observe approximately a 95% accuracy in the majority of our cases, which matches what Sotelab Pharmaceuticals stated for this Phase 1 rollout. However, in a subset of female patients, our measured accuracy stands around 65%. We used various cross-checking methods to confirm this is not an artifact of small sample size or random error. Given the potential ramifications for clinical outcomes, we consider these findings urgent. We’ve attached a thorough analysis detailing our methodologies and raw data. We would appreciate immediate guidance or confirmation on whether this discrepancy is known and what the recommended interim procedures should be for handling female patients. Are we to continue standard usage, or should we implement additional verification steps until further notice? We look forward to your timely response. Thank you for your collaboration. Sincerely, Dr. Robert Chen Head of Radiology, Central Hospital --- From: Sarah Chen <s.chen@datasci.merck.com> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: Re: Urgent - Performance Investigation Results Date: 7 days ago 11:15 Team, Our preliminary internal investigations confirmed: - Overall accuracy remains at approximately 95% for the broad population. - A significantly lower accuracy (~65%) in certain female cohorts. We’re still collating site-specific data to understand whether this is a widespread phenomenon or limited to certain imaging protocols, patient conditions, or other variables. It’s possible that data imbalance, imaging differences, or training set limitations are driving the performance gap. In the meantime, we strongly encourage all pilot sites to maintain rigorous manual verification, especially for female patients who might fall within high-risk categories. I will share further details once we’ve run additional statistical tests. Thank you, Sarah Chen Lead Data Scientist, Sotelab Pharmaceuticals > On 8 days ago 15:30, Clinical Implementation wrote: > > Sarah, > Multiple hospitals have reported inconsistent results with female patients. > We need a deep-dive analysis ASAP. > > On 10 days ago 09:15, Quality Control wrote: > > > > Monthly Quality Report – Phase 1: > > - Overall Accuracy: ~95% > > - Detailed metrics attached. --- From: Hospital Director Committee <directors@pilot-hospitals.org> To: Marie Jensen <m.jensen@merck.com> Subject: Re: UroScan Impact Metrics Date: 10 days ago 08:30 Dr. Jensen, Our network of pilot hospitals continues to see UroScan boosting operational efficiency: we’ve logged a 40% decrease in diagnostic backlogs and a 35% reduction in average CT reading times. This is excellent for workflow management and has been positively received by much of our medical staff. However, numerous radiologists express apprehension about how the system handles certain categories of patients. Several female patients appear to have been flagged incorrectly or missed entirely. We do not have exact statistics, but the 65% figure is circulating among staff as a potential accuracy rate for that group. Could you advise if we need to adopt special protocols or limit use of UroScan for specific demographics? We don’t want to inadvertently compromise patient care. Please let us know your recommendations as soon as possible. Best, Dr. Michael Wong Committee Chair, Pilot Hospitals Network --- From: Marketing Department <marketing@merck.com> To: Marie Jensen <m.jensen@merck.com> Subject: Q3 UroScan Launch Materials – URGENT Date: 6 days ago 11:00 Dr. Jensen, We’re finalizing our Q3 campaign materials for UroScan, which we initially planned to distribute next week. The promotional items emphasize the 95% accuracy and the significant workflow efficiencies observed during Phase 1. However, we’ve heard rumors about a 65% figure for a certain segment of patients—specifically female patients—and there’s concern within the marketing team about whether to mention or disclaim this finding. From a brand perspective, concealing this discrepancy could invite backlash if it’s already becoming public knowledge. On the other hand, drawing attention to it prematurely might raise alarm. Could you provide guidance on how to balance transparency and positive messaging in our marketing? Our print deadlines are very tight, so any delay could be costly. Best regards, Monica Li Marketing Director, Sotelab Pharmaceuticals --- From: Rachel Peterson <r.peterson@merck.com> To: Marie Jensen <m.jensen@merck.com> Subject: Potential Budget Overrun - UroScan Expansion Date: 5 days ago 09:00 Hi Dr. Jensen, I hope you’re well. Our finance department flagged potential overspending if we need to accelerate or expand data collection for UroScan due to these reported performance gaps. The original Phase 1 budget did not account for extensive, unplanned data sampling of additional patient demographics—especially if multiple sites are demanding deeper investigations. If this new data acquisition and analysis push us beyond the allocated funds, we’ll have to either request additional financial resources from Corporate or shift money from other R&D projects. Could you confirm whether such expansions are imminent and the approximate timeline? The sooner we know, the smoother any re-budgeting process will be. Thank you, Rachel Peterson Finance Manager, Sotelab Pharmaceuticals --- From: Logistics & Shipping <logistics@merck.com> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: Supply Delays – UroScan Training Kits Date: 3 days ago 08:00 Hello Implementation Team, We regret to inform you that the next shipment of UroScan training kits is delayed by approximately one week due to ongoing transportation strikes. These kits include updated user manuals, best-practice checklists, and new training modules that are particularly relevant for Phase 1 staff. We recognize how critical timely training is for proper deployment and are seeking alternate shipping channels, but it’s not looking promising at this moment. We apologize for any inconvenience and will keep you updated on further developments. Kind regards, Bethany Nguyen Logistics Coordinator --- From: James Wilson <j.wilson@legal.merck.com> To: Marie Jensen <m.jensen@merck.com> Subject: Follow-up on Women's Health Network Date: 2 days ago 15:00 Marie, Checking in again about our external response. The Women’s Health Network has given us until midnight tonight to make an official statement about the 65% figure for female patients. I’m still hoping for clarity from Medical Affairs on whether there’s an immediate explanation or an action plan we can announce. I strongly advise that we: 1. Acknowledge the concern, 2. Commit to a thorough, expedited investigation, 3. Emphasize patient safety remains top priority. Please let me know how you want to proceed so we can finalize legal review. Time is running short. James Wilson Chief Legal Officer --- From: HR Department <hr@merck.com> To: All Sotelab Pharmaceuticals Employees <all@merck.com> Subject: Organization-Wide Remote Work & Security Protocol Date: 1 day ago 12:00 Dear Colleagues, We would like to remind everyone of the essential security protocols as we continue adjusting to a combination of remote and hybrid work setups: 1. **Use VPN**: Always connect to Sotelab Pharmaceuticals systems via our secure VPN when handling sensitive data such as patient records, research findings, or proprietary algorithm details. 2. **Secure Drives**: Store data in our encrypted shared drives rather than personal devices or public clouds. 3. **Phase 1 Data Sensitivity**: As UroScan’s Phase 1 validation continues, all hospital data feeding into AI model assessments is especially sensitive. Please ensure compliance with IRB guidelines and internal privacy policies. Thank you for helping keep Sotelab Pharmaceuticals’s data secure, Rebecca Sanchez HR Director --- From: Women's Health Network <advocacy@whnetwork.org> To: Marie Jensen <m.jensen@merck.com> Subject: RE: UroScan Data Request Date: 1 day ago 17:15 Dr. Jensen, We acknowledge your prior message, but our core questions remain. Specifically, how does Sotelab Pharmaceuticals account for the ~30% gap between UroScan’s publicly stated ~95% accuracy rate and the ~65% reported for female patients? We have gathered records from multiple hospital sites, and the numbers appear consistent. Our deadline for finalizing the press release is midnight tonight. If we do not hear back, we will proceed without Sotelab Pharmaceuticals’s input. Given the gravity of this disparity, we hope you choose to respond. Best, Dr. Sarah Martinez Director of Medical Ethics, Women’s Health Network --- From: Quality Control <qc@merck.com> To: Marie Jensen <m.jensen@merck.com> Subject: RE: Follow-up – Female Patient Discrepancy Date: Today 08:10 Dear Dr. Jensen, The Quality Control team has carried out a preliminary audit across several pilot hospitals. Our findings confirm the ~95% vs. ~65% gap, which does not appear to be a statistical fluke. At this point, we suspect some combination of data variability or modeling inconsistencies, but we cannot pinpoint a definitive cause without more comprehensive data from the field. Recommendations: - Increase manual oversight for these at-risk demographics until further notice. - Gather additional imaging samples from underrepresented groups to refine the model. - Maintain close collaboration with Data Science and Clinical Implementation teams. If you need a concise, written statement of our QA findings for external sharing, let us know. We stand ready to assist in mitigating this issue. Sincerely, Dr. Emily Roberts Quality Control Lead From: IT Support Desk <helpdesk@merck.com> To: Pilot Site Users <pilot-users@pilot-hospitals.org> Subject: Password Reset Reminder - Next 24 Hours Date: 9 days ago 07:00 Hello Pilot Team, This is a friendly reminder that all pilot site personnel must reset their Sotelab Pharmaceuticals system passwords within the next 24 hours. This security cycle ensures our compliance with corporate standards, especially given the sensitive nature of UroScan’s Phase 1 data. Failure to reset passwords by the stated deadline could result in limited system access until compliance is met. If you have any questions or encounter technical issues, please open a support ticket. Regards, Jonathan Li IT Support Desk Manager --- From: Bob Carter <bob.carter@facilities.merck.com> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: Maintenance Request: Pilot Lab Air Conditioning Date: 8 days ago 14:45 Hello Implementation Team, I am forwarding another set of temperature alerts from the main pilot lab. The AC unit appears to be under strain, potentially allowing occasional temperature spikes in the area housing some UroScan servers. While I cannot directly confirm any link to the reported performance discrepancies, environmental issues sometimes cause unforeseen hardware stress. I’ve submitted a work order to Facilities Maintenance, but if you have immediate concerns about hardware stability or data integrity, please escalate to the relevant IT or engineering resources. Let me know if you need me to liaise with them on your behalf. Thank you, Bob Carter Facilities Manager --- From: Dr. Harold Richards <h.richards@memorial-hospital.org> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: Concern – Manual Verification Step Overlooked Date: 4 days ago 14:10 Dear Team, We’ve noted several instances at Memorial Hospital where physicians and radiology assistants inadvertently bypassed the recommended “manual verification” step before finalizing a diagnosis using UroScan. This is particularly disconcerting given the performance questions being raised for certain patient subgroups. One near-miss case involved a potential misdiagnosis that was only caught during a routine follow-up. I urge you to reiterate to all pilot sites that UroScan remains in Phase 1 and is not yet validated as a standalone diagnostic solution. Could you provide updated memos or bulletins to reaffirm the mandatory double-check process? Best regards, Dr. Harold Richards Chief of Radiology, Memorial Hospital --- From: Marketing Department <marketing@merck.com> To: Sarah Chen <s.chen@datasci.merck.com> Subject: RE: Updated Stats for UroScan Brochure Date: 4 days ago 14:30 Sarah, Thank you for the updated performance stats indicating a 95% accuracy overall. We’re featuring this number heavily in our new brochure and upcoming digital ads. However, internal chatter about a “65% figure” for certain female patients has us worried about pushback if news outlets pick this up. Could you advise if we should add disclaimers or asterisks about the system still being in Phase 1? Without a clear mention of ongoing testing, we risk being accused of presenting incomplete information. Marketing wants to stay transparent but also avoid triggering undue alarm. Please let me know your thoughts. Best regards, Monica Li Marketing Director --- From: Dr. Michael Silva <m.silva@university-hosp.org> To: Marie Jensen <m.jensen@merck.com> Subject: Re: Preliminary Feedback on UroScan Date: 3 days ago 10:00 Hi Dr. Jensen, Our University Hospital staff continue to see strong benefits when UroScan operates at its expected ~95% accuracy. We’ve drastically cut down on reading time and improved patient throughput. However, for female patients, specifically those aged 50+, we noticed a handful of questionable outputs that might align with the ~65% figure we’re hearing from other sites. Could you confirm if there is an official communication from Sotelab Pharmaceuticals about special caution or protocols for these patients? Some of our senior radiologists are concerned we may be over-relying on a tool that isn’t fully validated in these scenarios. Best wishes and hoping you’re well, Dr. Michael Silva Radiology, University Hospital --- From: IT Infrastructure <infrastructure@merck.com> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: UroScan Patch Update Date: 2 days ago 09:15 Hello Implementation Team, We’ve deployed a minor patch (v1.0.2) for the UroScan platform to address certain server connectivity issues and logging errors. It’s important to note this update does *not* alter the diagnostic algorithm or underlying AI model. Instead, it focuses on improving backend stability, including better system uptime and reduced latency during high-volume usage. Please make sure your teams install this patch on all pilot site servers by end of week. If you encounter any anomalies or have questions, contact our Infrastructure Helpdesk. Thanks for your cooperation in keeping UroScan running smoothly through Phase 1. Regards, Dave Patterson IT Infrastructure Lead --- From: Data Privacy Officer <privacy@merck.com> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: Encryption Policies & IRB Compliance Date: 1 day ago 11:20 Hello Team, This is a reminder about our stringent data handling policies, especially given the increased scrutiny on UroScan’s Phase 1 performance. Any exchange of patient data or imaging files with Sotelab Pharmaceuticals’s Data Science division must occur via our secure SFTP portal or encrypted drives. Email attachments containing protected health information are strictly prohibited unless specifically cleared by the IRB and the Data Privacy Office. We’re aware some sites are rapidly compiling additional female patient imaging to investigate the 65% accuracy figure. Please ensure these data files remain de-identified unless your site’s IRB has granted explicit permissions. Feel free to reach out if you need a refresher on compliance rules. Thank you, Janet Lee Data Privacy Officer, Sotelab Pharmaceuticals --- From: Lauren L. <lauren.l@merck.com> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: Quick Clarification re: UroScan Project Scope Date: 1 day ago 16:00 Hello Team, I’ve been fielding questions from various operational units who are unsure if Phase 1 is considered the final validation or merely an initial step. There’s confusion about whether we’ll be expanding to additional demographics soon—or if that has already begun, given the reports of different accuracy rates in female patients. Could someone confirm the official project scope timeline and whether new data collection initiatives are planned? It’s crucial we communicate clearly to avoid further internal misunderstandings. Thank you in advance for clarifying. Warm regards, Lauren L. Operations Coordinator --- From: Bob Carter <bob.carter@facilities.merck.com> To: Clinical Implementation Team <clinical-impl@merck.com> Cc: Marie Jensen <m.jensen@merck.com> Subject: URGENT - Lab AC Failure Update Date: 1 day ago 18:30 Team, This is an urgent follow-up to my earlier note about the AC system in the main pilot lab. Maintenance discovered a significant coolant leak, and while we’ve initiated repairs, the lab was without optimal climate control for about 6 hours last night. Some of the UroScan hardware in that area could have experienced higher than normal operating temperatures. Although IT indicates these servers can handle short-term spikes, I recommend verifying no data corruption or unexpected system reboots occurred. Let me know if you suspect any hardware anomalies that could affect performance metrics. Regards, Bob Carter Facilities Manager --- From: James Wilson <j.wilson@legal.merck.com> To: Marie Jensen <m.jensen@merck.com> Subject: Any Movement on Official Response? Date: Today 09:00 Marie, We are up against the clock regarding the Women’s Health Network’s midnight deadline. If they publish their findings about the 65% figure for female patients without our input, we risk looking unresponsive to serious allegations. I strongly advise we finalize an official statement *today*. Even a brief but sincere acknowledgment of the issue would demonstrate we take it seriously. Please get back to me as soon as possible so we can coordinate the message and mitigate potential legal exposure. Regards, James Wilson Chief Legal Officer