From dementia screening to readmission prediction in hospitalized patients: Evaluation and development of diagnostic and prognostic instruments

From dementia screening to readmission prediction in hospitalized patients: Evaluation and development of diagnostic and prognostic instruments Cumulative thesis submitted for the degree of Dr. sc. in Health Sciences at the Department of Health Sciences and Medicine, University of Lucerne Hwang, A.B., Schuepfer, G., Nyffeler, T., and Boes, S. (2019). Validity of screening instruments for the detection of dementia and mild cognitive impairment in hospital inpatients: A systematic review of diagnostic accuracy studies. PLOS One, 14(7): e0219569. https://doi.org/10.1371/journal.pone.0219569 Hwang, A.B., Schuepfer, G., Pietrini, M., and Boes, S. (2021). External validation of EPIC’s Risk of Unplanned Readmission model, the LACE+ index and SQLape as predictors of unplanned hospital readmissions: A monocentric, retrospective diagnostic cohort study in Switzerland. PLOS One, 16(11) : e0258338. https://doi.org/10.1371/journal.pone.0258338 Havranek, M., M., Hwang, A.B., Funke, I., Kuhlen, D., Liedtke, D., and Boes, S. (2025). Machine learning predictions of unplanned readmissions using electronic medical records: Predictor importance across medical and surgical patient populations. Submitted to PLOS One Dec. 2024 presented by Aljoscha B. Hwang 06-203-004 Lucerne, 2025 DOI: 10.5281/zenodo.14537199 I Acknowledgements Writing this cumulative dissertation has been a journey marked by both challenges and meaningful milestones. I would like to take this opportunity to express my sincere gratitude to the many people who have supported, guided, and encouraged me throughout this process. I would like to thank my supervisor, Stefan Boes, for accepting me as his external doctoral student at the University of Lucerne and for his scientific advice. Thanks also to Armin Gemperli, who kindly agreed to be the second reviewer of this thesis. I am particularly grateful to Guido Schuepfer, former Chief Clinical Officer of the Cantonal Hospital Lucerne (LUKS), for his valuable guidance and scientific support throughout my employment at the LUKS. Only with his support was I able to prospectively collect inpatient data and become familiar with EPIC’s state-of-the-art electronic medical record system and predictive analytics module. My deepest gratitude also goes to Daniel Liedtke, Chief Executive Officer of the Hirslanden Group, for his interest in my work, his invaluable contributions, and his encouragement to persevere. Thank you, Dani! Further, I am grateful to all my co-authors for their contributions, in particular, their thoughtful guidance and constructive feedback. Special thanks to Michael Havranek, Research Director of the Competence Center for Health Data Science at the University of Lucerne, for the outstanding cooperation. Your support and professionalism have made our collaboration both successful and enjoyable. Last but not least, I want to thank my parents, Imsook and Juergen Hwang, and my close friend Jan Rheinberger for their love and support, and for always believing in me and being there for me. II Short summary Switzerland’s healthcare system is among the best but faces several challenges, such as balancing quality, affordability, and efficiency. Early detection of inpatients with cognitive impairment and those at risk for unplanned hospital readmissions, in combination with effective targeted interventions, can significantly contribute to high-quality and financially sustainable care. When considering the application of screening tests and risk prediction tools, understanding their validity in the target population and setting in which implementation is intended is crucial. This thesis critically appraises existing tests and tools, providing decision-makers with necessary insights. Furthermore, it presents the development of customized machine learning models tailored to local idiosyncrasies. The thesis consists of three studies. The first study systematically reviews and evaluates the diagnostic accuracy of screening tests used to detect dementia and mild cognitive impairment in hospitalized patients. Only a small number of tests are found, with no single best test for use in hospital settings. The second study externally validates a widely used international risk prediction tool for unplanned hospital readmissions, intended for implementation at a tertiary care provider in Central Switzerland. The model’s performance appears less favorable compared to two alternative models and its originally reported performance in the development study. Updating the model appears to be indicated. The third study builds upon the findings of the second study and constructs machine learning models to predict unplanned hospital readmissions, adapted from the definition provided by the American Centers for Medicare and Medicaid Services (CMS). Results show that unplanned readmissions within 30 days can be accurately predicted using only predischarge EMR data to inform clinical decision-making. III Abbreviations 6-CIT Six Item Cognitive Impairment Test AM Ante Meridiem / Before noon AMTS Abbreviated Mental Test Score ANQ Swiss National Association for Quality Development in Hospitals and Clinics AI Artificial Intelligence AUC Area Under the Curve CDT Clock-Drawing Test CMS Centers for Medicare & Medicaid Services CPS Cognitive Performance Scale EHR Electronic Health Record EMR Electronic Medical Record GDP Gross Domestic Product LACE+ Length of stay, Acuity of admission, Comorbidities, Emergency department use LLM Large Language Model LUKS Cantonal Hospital Lucerne MCI Mild Cognitive Impairment MMSE Mini-Mental-Status-Exam NCDs Noncommunicable Diseases OECD Organization for Economic Cooperation and Development PICO Population, Intervention, Comparison and Outcome SD Standard Deviation SQLape Striving for high Quality Level and Alleviation of Patients Expenditures T&C Time and Change test WHO World Health Organization 1 Chapter 1 Introduction 1.1 Background The Swiss healthcare system is recognized for its high quality, as substantiated by a comprehensive analysis conducted by the Organization for Economic Cooperation and Development (OECD) and World Health Organization (WHO) in October 2011 [1], the high levels of satisfaction expressed by the Swiss population [2, 3], and other international comparative studies [4, 5]. Key strengths include the guaranteed access to healthcare and a comprehensive range of medical services covered by mandatory health insurance, both of which contribute to a high quality of life and an above-average life expectancy [6]. However, the Swiss system is not without its problems. Notable weaknesses include limited transparency, a lack of targeted management, incomplete statistical and analytics fundamentals, and misaligned incentives, all of which contribute to suboptimal resource utilization and unnecessary costs [1]. According to the Federal Council, Switzerland’s highest executive authority, four critical health policy challenges must be addressed over the next decade, three of which are relevant to this thesis. First, Switzerland is experiencing significant demographic and societal changes, characterized by an ageing population, declining birth rates, and sustained high levels of immigration [7]. The increasing proportion of older adults is associated with a higher prevalence of age-related conditions, particularly noncommunicable disease (NCDs) [8]. These conditions not only impose a substantial burden on individuals but also place considerable strain on the healthcare system [9]. Second, among OECD countries, Switzerland has one of the most expensive healthcare systems. In 2023, total nominal healthcare expenditure rose by 3.8%, exceeding the 2.5% growth recorded in the previous year (2022). In relation to the gross domestic product (GDP), this corresponds to 11.8%, up from 11.6% in the previous year [10]. 2 This means, healthcare expenditure relative to the GDP has more than doubled over the last half-century or so [11]. From a macroeconomic standpoint, this trend is not problematic per se in an ageing society. However, a portion of these costs arises from inadequate quality of care associated with adverse events and unfavorable outcomes [12, 13]. Third, despite the significant potential of digital solutions to improve health outcomes, treatment quality, and efficiency, digitization in healthcare has progressed more slowly than in some neighboring countries [14]. A recent study assessed the potential impact of healthcare digitization in Switzerland, estimating a possible reduction in healthcare expenditures of up to CHF 8.2 billion, equivalent to 11.8% of total addressable healthcare costs [15]. The outlined health policy challenges emphasize the need for targeted interventions to address the increasing prevalence of age-related conditions, particularly NCDs, and their associated burden. Health promotion campaigns, disease prevention efforts, and early detection initiatives are called for to mitigate the impact of non-communicable diseases by fostering healthier lifestyles and enabling timely medical interventions [12, 13]. Furthermore, a stronger emphasis on patient-centered approaches, with interventions provided according to evidence-based guidelines, alongside leveraging digital solutions, is essential to efficiently address persistent gaps in quality of care [12]. 1.2 Rational In the context of inpatient care, early stratification of inpatients based on their specific risks and needs is crucial for delivering high-quality and financially sustainable care. Recent studies and reports have identified certain non-communicable diseases exhibiting particularly rapid growth rates, with dementia being a prominent example. According to Alzheimer Europe, the number of people with dementia in Europe is projected to nearly double by 2050, reaching 14,298,671 in the European Union [16]. In Switzerland, approximately 156,900 individuals are affected by dementia as of 2024, with a prevalence of up to 40% among hospital inpatients [17-19]. Despite the lack of a cure, early diagnosis is emphasized in many national dementia strategies due to its potential benefits, including improved symptom management, delayed institutionalization, and the facilitation of coordinated care plans, among other 3 advantages [20, 21]. In the inpatient setting, a known dementia diagnosis can facilitate the provision of more personalized and appropriate care, including decisions regarding anesthesia, medication management, and tailored discharge planning. This, in turn, helps reduce the risk of adverse events and unfavorable outcomes, which are more likely among inpatients with dementia [22-25]. Studies suggest, that approximately one in ten inpatients experiences at least one adverse event during hospitalization, while every second event is considered potentially preventable. [26, 27]. Furthermore, it is important to note that patient safety-related adverse events, in general, not only cause harm and suffering but also have significant economic implications. In 2019, total excess costs amounted to an estimated CHF 347 million, corresponding to 2.2% of Switzerland’s annual inpatient expenditures [28]. Among the various indicators of unfavorable outcomes in inpatient care, unplanned readmissions stand out as a widely accepted and commonly used measure of quality [29-32]. As the population ages and the proportion of elderly, frail inpatients increases, ensuring continuity of care becomes essential for achieving positive outcomes. Accordingly, unplanned readmissions are assessed annually by the Swiss National Association for Quality Development in Hospitals and Clinics (ANQ) as part of its national quality monitoring program. In many developed countries, including Switzerland, rates of unplanned readmissions range from 5% to 20% [33-39]. Notably, a substantial proportion of these readmissions result from preventable factors such as inadequate discharge planning, unresolved medical issues, or insufficient follow-up care [40]. A systematic review and meta-analysis found that approximately 23% of readmissions are deemed avoidable [41, 42]. Consequently, the costs associated with unplanned readmissions are significant. While Swiss-specific cost data is unavailable, the cost of unplanned readmissions for Medicare patients alone stands at an estimated $26 billion annually, out of which $17 billion are potentially preventable [43, 44]. These considerations underscore the importance of implementing and utilizing instruments specifically designed to identify inpatients with dementia and those at risk for unplanned readmissions, thereby facilitating individualized stratification based on inpatients’ needs and risks. In this context, screening tests and risk prediction tools have emerged as valuable instruments, leveraging patient-level data to guide targeted interventions aimed at reducing the probability of adverse events and unplanned 4 readmissions. However, many of these instruments lack generalizability and transportability, making them unsuitable for direct implementation in clinical practice without prior critical appraisal. 1.3 Objectives of this thesis This thesis aims to contribute to the overarching goal of maintaining high-quality and financially sustainable healthcare. Its main contribution lies in the identification, validation, and development of instruments to detect hospital patients with conditions or characteristics that predispose them to adverse events and unfavorable outcomes during or after hospitalization. By implementing these instruments in combination with effective (preventive) measures, healthcare providers will be better equipped to improve the quality of care and reduce costs. The three studies presented in this dissertation cover the following topics: 1. Paper-based screening instruments for the detection of cognitive impairment in hospital inpatients: The study entitled: “Validity of screening instruments for the detection of dementia and mild cognitive impairment in hospital inpatients: A systematic review of diagnostic accuracy studies” aims to provide clinicians, who wish to promote early detection of cognitive impairment by implementing targeted screening, an up-to-date choice of cognitive tests with the most extensive evidence base for the use in elderly elective hospital inpatients (chapter 2). 2. External validation and comparison of an Electronic Health Record System (EHRs) embedded decision support model: The study entitled “External validation of EPIC’s Risk of Unplanned Readmission model, the LACE+ index and SQLape as predictors of unplanned hospital readmissions: A monocentric, retrospective, diagnostic cohort study in Switzerland” aims to provide critical evidence on the generalizability and transportability of unplanned readmission prediction models at the largest tertiary healthcare provider in Central Switzerland (chapter 3). 3. Development of a Swiss clinical decision-support model for unplanned readmissions: In the study entitled “Machine learning predictions of unplanned readmissions using electronic medical records: Predictor importance across medical and surgical patient populations” we investigate whether unplanned readmissions 5 can be accurately predicted using solely predischarge data from Electronic Medical Records (EMRs). This research is conducted in collaboration with a leading private hospital group, including eight general hospitals located in German-speaking cantons of Switzerland (chapter 4). 1.4 The process of identifying and evaluating clinical instruments and tools A substantial portion of clinical practice relies on instruments designed to synthesize clinical patient characteristics with the existing evidence base to guide decision-making. Among others, these include screening tests, diagnostic tests, and risk prediction tools, each tailored to specific use cases. Screening tests are typically used to detect potential health conditions early and are designed for broad preventive application [45]. In contrast, diagnostic tests are confirmatory, often more complex and invasive, and aim to establish a definitive diagnosis [46]. Risk prediction tools, on the other hand, are based on probabilistic models designed to forecast future risk or outcomes [47]. All of them leverage patient-specific data or biospecimens, medical knowledge, and statistical or algorithmic approaches to support (proactive) clinical decision-making. Selecting the most appropriate screening test or risk prediction tool for a specific use case is a complex and challenging task, not only due to the abundance of available alternatives but also because of the lack of transparent and meaningful reporting [48, 49]. Utilizing the PICO framework (Population, Intervention, Comparator, Outcome) can help structure and streamline the literature search process, facilitating a focused and systematic approach to identify relevant studies for critical appraisal [50]. Screening tests and risk prediction tools are typically evaluated in the context of diagnostic accuracy studies and prognostic accuracy studies, respectively. Very much like other clinical studies, diagnostic and prognostic accuracy studies are susceptible to bias due to shortcomings in design and conduct [51-53]. Furthermore, diagnostic and prognostic accuracies are not fixed properties of an instrument but dependent on setting and patient characteristics. For instance, sensitivity of a screening test may vary significantly depending on whether it is utilized in a primary or tertiary care setting. Both risk of bias and concerns regarding applicability can be systematically evaluated using quality assessment tools, provided that the necessary information is reported [54, 55]. 12 References [1] OECD-Berichte über Gesundheitssysteme: Schweiz [Internet]. OECD-WHO; 2011 [cited 2024 Dec 31]. Available from: https://www.oecd.org/de/publications/2011/10/oecd-reviews-of-health-systemsswitzerland-2011_g1g14ce1.html. [2] Abstimmungen zum Gesundheitswesen [Internet]. Bundesamt für Gesundheit BAG; 2024 [cited 2024 Dec 31]. Available from: https://www.bag.admin.ch/bag/de/home/strategie-undpolitik/abstimmungen.html#:~:text=Die%20Volksabstimmung%20zur%20%C3%8 4nderung%20des,von%2053%2C31%20Prozent%20angenommen. [3] IHP-Befragungen: Wohnbevölkerung ab 18 Jahren [Internet]. Bundesamt für Gesundheit BAG; 2024 [cited 2024 Dec 31]. Available from: https://www.bag.admin.ch/bag/de/home/dasbag/publikationen/forschungsberichte/forschungsberichte-international-healthpolicy-survey-ihp-des-commonwealth-fund-laendervergleiche/ihp-befragungenwohnbevoelkerung-ab-18-jahren.html. [4] Schoen C, Osborn R, Squires D, Doty MM, Pierson R, Applebaum S. How health insurance design affects access to care and costs, by income, in eleven countries. Health affairs. 2010;29(12):2323-34. [5] Savidan A, Cathieni F, Carron T, Perriraz M, Storari C, Gilles I, et al. Brief survey on current Swiss hospitals satisfaction practices & Literature review on international large-scale initiatives evaluating inpatient satisfaction. 2022. [6] OECD Better Life Index - Switzerland [Internet]. OECD; 2024 [cited 2025 Jan 1]. Available from: https://www.oecdbetterlifeindex.org/countries/switzerland/. [7] Schweiz Szenarien [Internet]. Bundesamt für Statistik; 2020 [cited 2025 Jan 2]. Available from: https://www.bfs.admin.ch/bfs/de/home/statistiken/bevoelkerung/zukuenftigeentwicklung/schweiz-szenarien.html. [8] Noncommunicable disease [Internet]. World Health Organization (WHO); 2023 [cited 2025 Jan 2]. Available from: https://www.who.int/news-room/factsheets/detail/noncommunicable-diseases?utm_source=chatgpt.com. [9] Armocida B, Monasta L, Sawyer S, Bustreo F, Segafredo G, Castelpietra G, et al. Burden of non-communicable diseases among adolescents aged 10–24 years in the EU, 1990–2019: a systematic analysis of the Global Burden of Diseases Study 2019. The Lancet Child & Adolescent Health. 2022;6(6):367-83. [10] Anderes M, Gersbach H. KOF-Prognose der Gesundheitsausgaben: Herbst 2024. KOF Studien; 2024. [11] Colombier C. Healthcare expenditure and fiscal sustainability: evidence from Switzerland. Public Sector Economics. 2018;42(3):279-301. [12] Bundesrat D. Die gesundheitspolitische Strategie des Bundesrates 2020-2030. Schweizerisches Eidgenossenschaft; 2019. [13] Mattig T, Chastonay P. Switzerland’s national non-communicable diseases strategy. Ann Chron Dise 2017; 2 (1). 2017;1002. [14] SmartHealthSystems - Digitalisierungsstrategien im internationalen Vergleich [Internet]. BertelsmannStiftung; 2018 [cited 2025 Jan 2]. Available from: 13 https://www.bertelsmannstiftung.de/de/publikationen/publikation/did/smarthealthsystems. [15] Digitization in healthcare: The CHF 8.2 billion opportunity for Switzerland [Internet]. McKinsey & Company and ETH Zürich; 2021 [cited 2025 Jan 2]. Available from: https://www.mckinsey.com/ch/~/media/mckinsey/locations/europe%20and%20 middle%20east/switzerland/our%20insights/digitization%20in%20healthcare/digi tization%20in%20healthcare%20%20the%20chf%2082%20billion%20opportunit y%20for%20switzerland%20en.pdf. [16] Europe A. Dementia in Europe Yearbook 2019: Estimating the prevalence of dementia in Europe. Alzheimer Europe. 2019;180. [17] Demenz in der Schweiz 2024: Zahlen und Fakten [Internet]. alzheimer Schweiz; 2024 [cited 2025 Jan 22]. Available from: https://www.alzheimerschweiz.ch/fileadmin/dam/Alzheimer_Schweiz/Dokumente/Ueber_Demenz/Zahle n-Fakten/Factsheet_DemenzCH_2024_DE.pdf. [18] Mukadam N, Sampson EL. A systematic review of the prevalence, associations and outcomes of dementia in older general hospital inpatients. International psychogeriatrics. 2011;23(3):344-55. [19] Reynish EL, Hapca SM, De Souza N, Cvoro V, Donnan PT, Guthrie B. Epidemiology and outcomes of people with dementia, delirium, and unspecified cognitive impairment in the general hospital: prospective cohort study of 10,014 admissions. BMC medicine. 2017;15:1-12. [20] Bassetti CL, Heldner MR, Adorjan K, Albanese E, Allali G, Arnold M, et al. The Swiss Brain Health Plan 2023–2033. Clinical and translational neuroscience. 2023;7(4):38. [21] Edick C, Holland N, Ashbourne J, Elliott J, Stolee P, editors. A review of Canadian and international dementia strategies. Healthcare management forum; 2017: SAGE Publications Sage CA: Los Angeles, CA. [22] Watkin L, Blanchard M, Tookman A, Sampson E. Prospective cohort study of adverse events in older people admitted to the acute general hospital: risk factors and the impact of dementia. International journal of geriatric psychiatry. 2012;27(1):76-82. [23] Bail K, Goss J, Draper B, Berry H, Karmel R, Gibson D. The cost of hospitalacquired complications for older people with and without dementia; a retrospective cohort study. BMC Health Services Research. 2015;15(1):91. [24] Catalán L, Oliveira D. P9: Preventable adverse events and related outcomes among people with dementia in hospital settings: Scoping review. International Psychogeriatrics. 2024;36(S1):122-3. [25] Chróinín DN, Deane V, Pulikotil Zachariah R, Stott K, Shepherd B, Perkins M, et al. The likelihood of hospital‐acquired complications in older people with dementia: a matched cohort study. Medical Journal of Australia. 2024;221(8):422-5. [26] Halfon P, Staines A, Burnand B. Adverse events related to hospital care: a retrospective medical records review in a Swiss hospital. International Journal for Quality in Health Care. 2017;29(4):527-33. [27] Schwendimann R, Blatter C, Dhaini S, Simon M, Ausserhofer D. The occurrence, types, consequences and preventability of in-hospital adverse events–a scoping review. BMC health services research. 2018;18:1-13. 14 [28] Giese A, Khanam R, Nghiem S, Staines A, Rosemann T, Boes S, et al. Assessing the excess costs of the in-hospital adverse events covered by the AHRQ’s Patient Safety Indicators in Switzerland. Plos one. 2024;19(2):e0285285. [29] van Walraven C. The utility of unplanned early hospital readmissions as a health care quality indicator. JAMA internal medicine. 2015;175(11):1812-4. [30] Fischer C, Lingsma HF, Marang-van de Mheen PJ, Kringos DS, Klazinga NS, Steyerberg EW. Is the readmission rate a valid quality indicator? A review of the evidence. PLoS One. 2014;9(11):e112282. [31] Hospital readmissions and socioeconomic deprivation in Switzerland [Internet]. Federal Office of Public Health FOPH; 2023 [cited 2025 Jan 21]. Available from: https://www.bag.admin.ch/dam/bag/en/dokumente/natgesundheitsstrategien/nat-programm-migration-undgesundheit/chancengleichheit-in-der-gesundheitsversorgung/zusammenfassungrehospitalisationen-undbenachteiligung.pdf.download.pdf/Abstract%20Hospital%20readmissions%20an d%20socioeconomic%20deprivation.pdf. [32] Rehospitalisationen [Internet]. ANQ; 2025 [cited 2025 Jan 21]. Available from: https://www.anq.ch/de/fachbereiche/akutsomatik/messinformationakutsomatik/rehospitalisationen/. [33] Amritphale A, Fonarow GC, Amritphale N, Omar B, Crook ED. All-cause unplanned readmissions in the United States: insights from the Nationwide Readmission Database. Internal Medicine Journal. 2023;53(2):262-70. [34] Tapking C, Boson AL, Rontoyanni VG, Kowalewski KF, Hundeshagen G, Haug VF, et al. Unplanned 30-day readmission rates after plastic and reconstructive surgery procedures: a systematic review and meta-analysis. European Journal of Plastic Surgery. 2020;43:701-12. [35] Magdelijns FJ, Schepers L, Pijpers E, Stehouwer CD, Stassen PM. Unplanned readmissions in younger and older adult patients: the role of healthcare-related adverse events. Eur J Med Res. 2016;21(1):35. [36] Wong EL, Cheung AW, Leung MC, Yam CH, Chan FW, Wong FY, et al. Unplanned readmission rates, length of hospital stay, mortality, and medical costs of ten common medical conditions: a retrospective analysis of Hong Kong hospital data. BMC health services research. 2011;11:1-8. [37] Stabile Rehospitalisationsraten im Jahr 2022 [Internet]. ANQ; 2024 [cited 2025 Jan 21]. Available from: https://www.anq.ch/de/medienmitteilungen/stabilerehospitalisationsraten-im-jahr2022/#:~:text=Heute%20publizierte%20der%20ANQ%20die,Rate%20dem%20Niv eau%20des%20Vorjahrs. [38] Shuster C, Hurlburt A, Tam P, Staples J. Unplanned hospital readmissions in British Columbia. 2018. [39] Gruneir A, Dhalla IA, van Walraven C, Fischer HD, Camacho X, Rochon PA, et al. Unplanned readmissions after hospital discharge among patients identified as being at high risk for readmission using a validated predictive algorithm. Open Med. 2011;5(2):e104-11. [40] Fluitman KS, van Galen LS, Merten H, Rombach SM, Brabrand M, Cooksley T, et al. Exploring the preventable causes of unplanned readmissions using root cause 15 analysis: Coordination of care is the weakest link. European Journal of Internal Medicine. 2016;30:18-24. [41] Van Walraven C, Bennett C, Jennings A, Austin PC, Forster AJ. Proportion of hospital readmissions deemed avoidable: a systematic review. Cmaj. 2011;183(7):E391-E402. [42] Avoidable Hospital Readmisison Literature Review on Australian and International Indicators [Internet]. Australian Commission on Safety and Quality in Health Care; 2019 [cited 2025 Jan 21]. Available from: https://www.safetyandquality.gov.au/publications-and-resources/resourcelibrary/avoidable-hospital-readmission-literature-review-australian-andinternational-indicators. [43] Goodman DC, Fisher ES, Chang C-H. After hospitalization: A Dartmouth Atlas report on readmissions among Medicare beneficiaries. The revolving door: A report on US hospital readmissions. 2013;2013:7-30. [44] Wan H, Zhang L, Witz S, Musselman KJ, Yi F, Mullen CJ, et al. A literature review of preventable hospital readmissions: Preceding the Readmissions Reduction Act. IIE Transactions on Healthcare Systems Engineering. 2016;6(4):193-211. [45] Earlier screening, risk assessment and health optimisation in perioperative pathways: guide for providers and integrated care boards [Internet]. NHS England; 2023 [cited 2025 Jan 11]. Available from: https://www.england.nhs.uk/longread/earlier-screening-risk-assessment-and-health-optimisation-in-perioperativepathways/. [46] Bolboacă SD. Medical Diagnostic Tests: A Review of Test Anatomy, Phases, and Statistical Treatment of Data. Comput Math Methods Med. 2019;2019:1891569. [47] Royston P, Moons KG, Altman DG, Vergouwe Y. Prognosis and prognostic research: developing a prognostic model. Bmj. 2009;338. [48] Collins GS, Dhiman P, Ma J, Schlussel MM, Archer L, Van Calster B, et al. Evaluation of clinical prediction models (part 1): from development to external validation. Bmj. 2024;384:e074819. [49] Kosack CS, Page AL, Klatser PR. A guide to aid the selection of diagnostic tests. Bull World Health Organ. 2017;95(9):639-45. [50] Hosseini M-S, Jahanshahlou F, Akbarzadeh MA, Zarei M, Vaez-Gharamaleki Y. Formulating research questions for evidence-based studies. Journal of Medicine, Surgery, and Public Health. 2024;2:100046. [51] Whiting PF, Rutjes AW, Westwood ME, Mallett S, Group Q-S. A systematic review classifies sources of bias and variation in diagnostic test accuracy studies. Journal of clinical epidemiology. 2013;66(10):1093-104. [52] Navarro CLA, Damen JA, van Smeden M, Takada T, Nijman SW, Dhiman P, et al. Systematic review identifies the design and methodological conduct of studies on machine learning-based prediction models. Journal of Clinical Epidemiology. 2023;154:8-22. [53] Dhiman P, Ma J, Andaur Navarro CL, Speich B, Bullock G, Damen JA, et al. Methodological conduct of prognostic prediction models developed using machine learning in oncology: a systematic review. BMC medical research methodology. 2022;22(1):101. 16 [54] Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Annals of internal medicine. 2011;155(8):529-36. [55] Lee J, Mulder F, Leeflang M, Wolff R, Whiting P, Bossuyt PM. QUAPAS: An Adaptation of the QUADAS-2 Tool to Assess Prognostic Accuracy Studies. Annals of Internal Medicine. 2022;175(7):1010-8. [56] Damen JA, Hooft L, Schuit E, Debray TP, Collins GS, Tzoulaki I, et al. Prediction models for cardiovascular disease risk in the general population: systematic review. bmj. 2016;353. [57] Bellou V, Belbasis L, Konstantinidis AK, Tzoulaki I, Evangelou E. Prognostic models for outcome prediction in patients with chronic obstructive pulmonary disease: systematic review and critical appraisal. Bmj. 2019;367. [58] Chan JY, Yau ST, Kwok TC, Tsoi KK. Diagnostic performance of digital cognitive tests for the identification of MCI and dementia: A systematic review. Ageing Research Reviews. 2021;72:101506. [59] Böger B, Fachi MM, Vilhena RO, Cobre AF, Tonin FS, Pontarolo R. Systematic review with meta-analysis of the accuracy of diagnostic tests for COVID-19. American journal of infection control. 2021;49(1):21-9. [60] Christopher O, Joe Z, Jessica M, Hutan A. Promising algorithms to perilous applications: a systematic review of risk stratification tools for predicting healthcare utilisation. BMJ Health & Care Informatics. 2024;31(1):e101065. [61] Riley RD, Archer L, Snell KIE, Ensor J, Dhiman P, Martin GP, et al. Evaluation of clinical prediction models (part 2): how to undertake an external validation study. BMJ. 2024;384:e074820. [62] Dhamnetiya D, Jha RP, Shalini S, Bhattacharyya K. How to Analyze the Diagnostic Performance of a New Test? Explained with Illustrations. J Lab Physicians. 2022;14(1):90-8. [63] Leeflang MMG, Allerberger F. How to: evaluate a diagnostic test. Clinical Microbiology and Infection. 2019;25(1):54-9. [64] Efthimiou O, Seo M, Chalkou K, Debray T, Egger M, Salanti G. Developing clinical prediction models: a step-by-step guide. BMJ. 2024;386:e078276. [65] Moons KG, Kengne AP, Grobbee DE, Royston P, Vergouwe Y, Altman DG, et al. Risk prediction models: II. External validation, model updating, and impact assessment. Heart. 2012;98(9):691-8. [66] Ramspek CL, Jager KJ, Dekker FW, Zoccali C, van Diepen M. External validation of prognostic models: what, why, how, when and where? Clinical Kidney Journal. 2020;14(1):49-58. [67] Cohen JF, Korevaar DA, Altman DG, Bruns DE, Gatsonis CA, Hooft L, et al. STARD 2015 guidelines for reporting diagnostic accuracy studies: explanation and elaboration. BMJ open. 2016;6(11):e012799. [68] Collins GS, Moons KG, Dhiman P, Riley RD, Beam AL, Van Calster B, et al. TRIPOD+ AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. bmj. 2024;385. [69] Karimi L, Mahboub-Ahari A, Jahangiry L, Sadeghi-Bazargani H, Farahbakhsh M. A systematic review and meta-analysis of studies on screening for mild cognitive impairment in primary healthcare. BMC Psychiatry. 2022;22(1):97. 17 [70] Yokomizo JE, Simon SS, Bottino CM. Cognitive screening for dementia in primary care: a systematic review. Int Psychogeriatr. 2014;26(11):1783-804. [71] Alzheimer-Demenz A, Cholinesterasehemmer C-H. Konsensus 2012 zur Diagnostik und Therapie von Demenz-kranken in der Schweiz. 2012. [72] Jackson TA, Naqvi SH, Sheehan B. Screening for dementia in general hospital inpatients: a systematic review and meta-analysis of available instruments. Age and ageing. 2013;42(6):689-95. [73] Zhou H, Della PR, Roberts P, Goh L, Dhaliwal SS. Utility of models to predict 28day or 30-day unplanned hospital readmissions: an updated systematic review. BMJ open. 2016;6(6):e011060. [74] Kansagara D, Englander H, Salanitro A, Kagen D, Theobald C, Freeman M, et al. Risk prediction models for hospital readmission: a systematic review. Jama. 2011;306(15):1688-98. [75] Van Calster B, Steyerberg EW, Wynants L, Van Smeden M. There is no such thing as a validated prediction model. BMC medicine. 2023;21(1):70. [76] Aubert CE, Folly A, Mancinetti M, Hayoz D, Donzé J. Prospective validation and adaptation of the HOSPITAL score to predict high risk of unplanned readmission of medical patients. Swiss medical weekly. 2016;146:w14335. [77] Brüngger B, Blozik E. Hospital readmission risk prediction based on claims data available at admission: a pilot study in Switzerland. BMJ Open. 2019;9(6):e028409. [78] Struja T, Baechli C, Koch D, Haubitz S, Eckart A, Kutz A, et al. What Are They Worth? Six 30-Day Readmission Risk Scores for Medical Inpatients Externally Validated in a Swiss Cohort. Journal of General Internal Medicine. 2020;35(7):2017-24. [79] Uhlmann M, Lécureux E, Griesser A-C, Duong HD, Lamy O. Prediction of potentially avoidable readmission risk in a division of general internal medicine. Swiss medical weekly. 2017;147. [80] Artetxe A, Beristain A, Graña M. Predictive models for hospital readmission risk: A systematic review of methods. Computer Methods and Programs in Biomedicine. 2018;164:49-64. [81] Mahmoudi E, Kamdar N, Kim N, Gonzales G, Singh K, Waljee AK. Use of electronic medical records in development and validation of risk prediction models of hospital readmission: systematic review. BMJ. 2020;369:m958. [82] Morgan DJ, Bame B, Zimand P, Dooley P, Thom KA, Harris AD, et al. Assessment of Machine Learning vs Standard Prediction Rules for Predicting Hospital Readmissions. JAMA Network Open. 2019;2(3):e190348-e. [83] Beecy AN, Gummalla M, Sholle E, Xu Z, Zhang Y, Michalak K, et al. Utilizing electronic health data and machine learning for the prediction of 30-day unplanned readmission or all-cause mortality in heart failure. Cardiovascular Digital Health Journal. 2020;1(2):71-9. [84] Huang Y, Talwar A, Chatterjee S, Aparasu RR. Application of machine learning in predicting hospital readmissions: a scoping review of the literature. BMC medical research methodology. 2021;21:1-14. 18 Chapter 2 Validity of screening instruments for the detection of dementia and mild cognitive impairment in hospital inpatients: A systematic review of diagnostic accuracy studies Joint work with Guido Schuepfer, Thomas Nyffeler, and Stefan Boes Published in PLOS ONE in 2019 https://doi.org/10.1371/journal.pone.0219569 19 Chapter 3 External validation of EPIC’s Risk of Unplanned Readmission model, the LACE+ index and SQLape as predictors for unplanned hospital readmissions: A monocentric, retrospective, diagnostic cohort study in Switzerland Joint work with Guido Schuepfer, Mario Pietrini, and Stefan Boes Published in PLOS ONE in 2021 https://doi.org/10.1371/journal.pone.0258338 20 Chapter 4 Machine learning predictions of unplanned readmissions using electronic medical records: Predictor importance across medical and surgical patient populations Joint work with Michael Havranek, Ilona Funke, Dominique Kuhlen, Daniel Liedtke, and Stefan Boes Submitted to PLOS ONE in 2024