FCFabienne Cotte

Fabienne Cotte

*This is an algorithmically generated author record based on publicly available information.

Fabienne Cotte has a total of 90 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    An AI-based symptom checker integrated into MomConnect maternal healthcare platform: a prospective, observational mixed-methods study

    by Marcel Schmude, Stephen Gilbert, Caroline Moor Larsson, Noxolo Gqada, Kris Lemon, Lauren Kotze, Reza Omar, Ebrahim January, Doreen Schlagenhauf, Hila Azadzoy, et al.

    High maternal mortality rates persist as a critical challenge in low-resource settings. Here, in this prospective, observational mixed-methods study, we evaluated a Diagnostic Decision Support System integrated into MomConnect, South Africa’s government-led WhatsApp-based maternal health platform. Between May and June 2023, 968 pregnant women and mothers were enrolled, of whom 86.3% faced multiple barriers to care. Behaviour was assessed via platform surveys and phone interviews at 1-week follow-up. A physician panel rated Diagnostic Decision Support System advice safe in 98.4% (181/184; P < 0.001) of cases. After use, 44.6% (266/596) changed care-seeking intentions, and uncertainty dropped from 17.3% to 6.0%. By follow-up, 43.0% (337/785) had sought care, up from 17.0% initially. Safe behaviours increased by 18.1% ( P = 0.001) and appropriate behaviour by 10.5% ( P = 0.053). Half of those receiving treatment had not initially planned to seek care. Integrating clinical artificial intelligence into a government health platform was associated with safe advice and improved health-seeking behaviour. ClinicalTrials.gov: NCT06790069; funded by the Rockefeller Foundation.


  • Open Access

    Appropriateness and utility of a clinical decision support system at the digital front door

    by Andreia Pimenta, Nisha Kini, Fabienne Cotte, Filipa Dias Lourenço, Miguel Paiva Pereira, Marcel Schmude, Athena Lemesiou, Stephen Gilbert, Tauseef Mehrali, Micaela Seemann Monteiro, et al.

    Digital front doors combining symptom assessment and self-triage are increasingly used, yet real-world evidence on their clinical integration remains limited. This prospective post-market clinical follow-up evaluated a clinical decision support system embedded in routine care within a large private healthcare network. Adults completed a structured symptom assessment before consultation, and both participants and physicians completed predefined questionnaires. Treating physicians generally judged the urgency advice and assessment report to be appropriate. When the report was reviewed before the consultation, physicians reported greater preparedness and perceived efficiency gains. Participants generally reported complete symptom capture and improved preparedness. These findings provide prospective real-world evidence that a symptom assessment CDSS can generate appropriate outputs in routine care and may support both patient preparation and clinician workflow when integrated into practice.

  • From Advice to Action — Real-World Behavior of Patients Using an Integrated Diagnostic Decision Support System for Navigating the Health Care System

    by Fabienne Cotte, Filipa Dias Lourenço, Miguel Paiva Pereira, Nisha Kini, Marcel Schmude, Andreia Pimenta, Athena Lemesiou, Stephen Gilbert, Tauseef Mehrali, Micaela Seemann Monteiro, et al.


  • Open Access

    Large language models for automatable real-world performance monitoring of diagnostic decision support systems: a comparison to manual doctor panel review in a prospective clinical study

    by Fabienne Cotte, Marcel Schmude, Philipp Bode, Oula Suliman, Filipa Dias Lourenço, Miguel Paiva Pereira, Nisha Kini, Vera Hartenstein, Allesandro Muscoloni, Lisa Stroux, et al.

    Background: Diagnostic decision support systems (DDSS) are increasingly deployed at scale, yet their diagnostic accuracy is insufficiently monitored once integrated into care. Traditional post-market surveillance relies on clinician review, which is costly, slow, and difficult to sustain. Large language models (LLMs) may offer a scalable and potentially automatable solution, but their performance in real-world monitoring remains unknown. Methods: We conducted a diagnostic accuracy substudy within ESSENCE, a prospective evaluation of Ada Health’s DDSS integrated into Portugal’s largest private healthcare network. Clinical notes and ICD-10 diagnoses from 498 encounters were anonymised and classified using a filter–map–match framework. Manual clinician review served as the reference standard. We compared eligibility classification and condition mapping between clinicians and GPT-4.1 and GPT-5, and assessed diagnostic accuracy of two DDSS versions using both reference sets. Findings: Manual review classified 385 of 498 encounters (77·3%) as eligible for diagnostic comparison. GPT-5 reproduced these classifications with 84·7% accuracy (κ=0·57), showing high sensitivity but only moderate specificity. Among 347 encounters judged eligible by both approaches, GPT-5 exactly matched clinician-assigned diagnoses in 93·6% and proposed clinically plausible alternatives in 3·5%. Diagnostic accuracy estimates based on manual versus GPT-5 mappings were statistically indistinguishable at Top-1 and Top-3 across the full analyzable sets, with one significant difference at Top-5. In the overlapping 346 cases, no statistical differences were observed. Across both reference sets, the experimental DDSS version outperformed the original only at the Top-5 threshold. Interpretation: LLMs can reproduce clinician review of real-world diagnostic encounters with close agreement. While GPT-5 performed comparably to clinicians for condition mapping, the eligibility filtering step - deciding which encounters should enter the diagnostic-accuracy analysis - remains the main source of divergence and is the priority for improvement. Embedding such approaches into health systems could enable automated and continuous performance and safety monitoring and support regulatory compliance. Broader evaluations across diverse care settings are needed to establish generalisability and equity impact. Funding: German Federal Ministry of Education and Research (NextGenerationEU, PATH project).


  • Open Access

    Medical Expert Knowledge Meets AI to Enhance Symptom Checker Performance for Rare Disease Identification in Fabry Disease: Mixed Methods Study

    by Anne Pankow, Nico Meißner-Bendzko, Jessica Kaufeld, Laura Fouquette, Fabienne Cotte, Stephen Gilbert, Ewelina Türk, Anibh Das, Christoph Terkamp, Gerhard-Rüdiger Burmester, et al.

    Background: Rare diseases, which affect millions of people worldwide, pose a major challenge, as it often takes years before an accurate diagnosis can be made. This delay results in substantial burdens for patients and health care systems, as misdiagnoses lead to inadequate treatment and increased costs. Artificial intelligence (AI)–powered symptom checkers (SCs) present an opportunity to flag rare diseases earlier in the diagnostic work-up. However, these tools are primarily based on published literature, which often contains incomplete data on rare diseases, resulting in compromised diagnostic accuracy. Integrating expert interview insights into SC models may enhance their performance, ensuring that rare diseases are considered sooner and diagnosed more accurately. Objective: The objectives of our study were to incorporate expert interview vignettes into AI-powered SCs, in addition to a traditional literature review, and to evaluate whether this novel approach improves diagnostic accuracy and user satisfaction for rare diseases, focusing on Fabry disease. Methods: This mixed methods prospective pilot study was conducted at Hannover Medical School, Germany. In the first phase, guided interviews were conducted with medical experts specialized in Fabry disease to create clinical vignettes that enriched the AI SC’s Fabry disease model. In the second phase, adult patients with a confirmed diagnosis of Fabry disease used both the original and optimized SC versions in a randomized order. The versions, containing either the original or the optimized Fabry disease model, were evaluated based on diagnostic accuracy and user satisfaction, which were assessed through questionnaires. Results: Three medical experts with extensive experience in lysosomal storage disorder Fabry disease contributed to the creation of 5 clinical vignettes, which were integrated into the AI-powered SC. The study compared the original and optimized SC versions in 6 patients with Fabry disease. The optimized version improved diagnostic accuracy, with Fabry disease identified as the top suggestion in 33% (2/6) of cases, compared to 17% (1/6) with the original model. Additionally, overall user satisfaction was higher for the optimized version, with participants rating it more favorably in terms of symptom coverage and completeness. Conclusions: This study demonstrates that integrating expert-derived clinical vignettes into AI-powered SCs can improve diagnostic accuracy and user satisfaction, particularly for rare diseases. The optimized SC version, which incorporated these vignettes, showed improved performance in identifying Fabry disease as a top diagnostic suggestion and received higher user satisfaction ratings compared to the original version. To fully realize the potential of this approach, it is crucial to include vignettes representing atypical presentations and to conduct larger-scale studies to validate these findings.


  • Open Access

    Holistic Human-Serving Digitization of Health Care Needs Integrated Automated System-Level Assessment Tools

    by Cindy Welzel, Fabienne Cotte, Magdalena Wekenborg, Baptiste Vasey, Peter McCulloch, Stephen Gilbert

    Digital health tools, platforms, and artificial intelligence– or machine learning–based clinical decision support systems are increasingly part of health delivery approaches, with an ever-greater degree of system interaction. Critical to the successful deployment of these tools is their functional integration into existing clinical routines and workflows. This depends on system interoperability and on intuitive and safe user interface design. The importance of minimizing emergent workflow stress through human factors research and purposeful design for integration cannot be overstated. Usability of tools in practice is as important as algorithm quality. Regulatory and health technology assessment frameworks recognize the importance of these factors to a certain extent, but their focus remains mainly on the individual product rather than on emergent system and workflow effects. The measurement of performance and user experience has so far been performed in ad hoc, nonstandardized ways by individual actors using their own evaluation approaches. We propose that a standard framework for system-level and holistic evaluation could be built into interacting digital systems to enable systematic and standardized system-wide, multiproduct, postmarket surveillance and technology assessment. Such a system could be made available to developers through regulatory or assessment bodies as an application programming interface and could be a requirement for digital tool certification, just as interoperability is. This would enable health systems and tool developers to collect system-level data directly from real device use cases, enabling the controlled and safe delivery of systematic quality assessment or improvement studies suitable for the complexity and interconnectedness of clinical workflows using developing digital health technologies.


  • Open Access

    Medical expert knowledge meets AI: Can expert knowledge improve symptom assessment apps - a new approach to rare diseases (Preprint)

    by Anne Pankow, Nico Meißner-Bendzko, Jessica Kaufeld, Laura Fouquette, Fabienne Cotte, Stephen Gilbert, Ewelina Türk, Anibh Das, Christoph Terkamp, Gerhard-Rüdiger Burmester, et al.

    BACKGROUND Although each individual disease in itself is rare, there are approximately 4 million people affected by rare diseases in Germany alone. Due to the limited knowledge, the time to diagnosis is often long and poses many challenges. Artificial intelligence (AI) approaches, like those used in symptom assessment applications (SAAs), can potentially help to detect rare diseases and thus shorten the time to diagnosis. Generally, SAAs use reasoning approaches applied to databases of both structured and unstructured medical knowledge from the medical literature. For rare diseases, however, there is only limited information available in medical literature. In this study we developed a new approach: abstracting medical expert knowledge from guided interviews and transforming them into clinical vignettes of the disease for the SAA knowledge base. OBJECTIVE Our goal was to integrate expert knowledge on the lysosomal storage diseases (LSDs) Fabry, Gaucher and Pompe disease into the Ada SAA, and to investigate whether the novel approach of guided interviews in combination with literature review leads to a better performance in symptom assessment than literature data sourcing alone. METHODS Clinical vignettes were created on the basis of interviews with 4 clinical experts from Hannover Medical School which gathered knowledge about typical symptom constellations. Structured case vignettes can easily be transcribed into the SAA’s knowledge base. Patients diagnosed with Fabry disease carried out an assessment using SAA versions generated from both the original disease modeling approach of data literature data sourcing alone and the novel approach. SAA performance and patient satisfaction were assessed using questionnaires. RESULTS With the help of the guided interviews 11 clinical vignettes were created to optimize the SAA’s LSD disease models. In the Fabry disease model comparison, participants generally favored the optimized SAA. With regard to diagnostic accuracy, there was a small difference between versions, favoring the optimized Fabry model. CONCLUSIONS The proposed novel approach for gaining medical knowledge about rare diseases for SAAs is a promising addition to the existing method of using literature research alone. Especially in case of rare diseases, knowledge is limited and medical literature is often incomplete – which often leads to diagnostic odysseys for the patients affected. Extending the traditional approach of disease modeling by extracting medical expert knowledge through clinical vignettes is a complementary method to fill and improve the knowledge base of SAAs, which may help to shorten the time to diagnosis for patients, including those suffering from rare diseases.

  • Open Access

    The Social Contract for Health and Wellness Data Sharing Needs a Trusted Standardized Consent

    by Stefanie Brückner, Toralf Kirsten, Peter Schwarz, Fabienne Cotte, Michael Tsesis, Stephen Gilbert


  • Open Access

    Exploring the Mental Health Characteristics of AI-Based Symptom Checker Users: A Comparison with the Global Burden of Disease Study (Preprint)

    by Oscar Freyer, François Bergey, Fabienne Cotte, Stephen Gilbert, Alicia Mehl

    BACKGROUND Mental health conditions pose a significant challenge for healthcare systems globally, and digital solutions have been proposed to address the challenges of limited access to healthcare, stigmatization, and lack of reliable data. An understanding of potential risk factors, comorbidities, and symptom constellations forms a necessary foundation for future digital solutions with diagnostic capabilities. Currently used digital products such as symptom checkers generate novel data sets that could provide insight into these correlations, helping both patients and healthcare providers. OBJECTIVE This study aimed to compare the characteristics of SC (Ada) users with a suggested mental health condition (MHC) in their symptom assessment (1) to all Ada users and (2) to the general population using the Global Burden of Diseases, Injuries, and Risk Factors (GBD) study 2019 dataset. METHODS Aggregated data from the SC was analyzed to provide a descriptive analysis of its users and compared to the World Population Prospects 2019 and the Global Burden of Disease study, which includes estimations for 195 countries and territories. Ada users worldwide who completed a symptom assessment between in 2020 or 2021 were included in the analysis. The study focused on user demographics, reported symptoms, and suggested conditions. RESULTS Out of 2,208,700 users, 20.9% received at least one MHC as the top suggestion. Female users (65.0%) were overrepresented in Ada's user base, and the largest number of users were aged 16-24 (57.0%). The average number of symptoms by users confirmed during the question flow was 6.7, with 1.9 symptoms entered initially. Major depressive disorder was the most frequently suggested MHC, affecting 10.2% of all Ada users, followed by other anxiety disorders (5.7%). Comparison of the Ada dataset with the GBD dataset showed that Depressive disorders and Anxiety disorders were the two most frequent MHCs in both datasets, with females more commonly affected than males. CONCLUSIONS Young and female users are overrepresented in the Ada userbase, especially among those with MHC suggestions compared with the GBD dataset. Ada's ability to reach these at-risk populations presents a significant opportunity for providing personalized and accessible healthcare solutions in the future. While the results cannot be easily generalized due to this population bias, this analysis highlights the potential of patient-reported data generated by symptom checkers as a source for epidemiological studies.

  • Open Access

    Holistic Human-Serving Digitization of Health Care Needs Integrated Automated System-Level Assessment Tools (Preprint)

    by Cindy Welzel, Fabienne Cotte, Magdalena Wekenborg, Baptiste Vasey, Peter McCulloch, Stephen Gilbert

    UNSTRUCTURED Digital health tools, platforms, and artificial intelligence– or machine learning–based clinical decision support systems are increasingly part of health delivery approaches, with an ever-greater degree of system interaction. Critical to the successful deployment of these tools is their functional integration into existing clinical routines and workflows. This depends on system interoperability and on intuitive and safe user interface design. The importance of minimizing emergent workflow stress through human factors research and purposeful design for integration cannot be overstated. Usability of tools in practice is as important as algorithm quality. Regulatory and health technology assessment frameworks recognize the importance of these factors to a certain extent, but their focus remains mainly on the individual product rather than on emergent system and workflow effects. The measurement of performance and user experience has so far been performed in ad hoc, nonstandardized ways by individual actors using their own evaluation approaches. We propose that a standard framework for system-level and holistic evaluation could be built into interacting digital systems to enable systematic and standardized system-wide, multiproduct, postmarket surveillance and technology assessment. Such a system could be made available to developers through regulatory or assessment bodies as an application programming interface and could be a requirement for digital tool certification, just as interoperability is. This would enable health systems and tool developers to collect system-level data directly from real device use cases, enabling the controlled and safe delivery of systematic quality assessment or improvement studies suitable for the complexity and interconnectedness of clinical workflows using developing digital health technologies.