Psychiatric Annals

CME Article 

Early Screening in the Emergency Department for Posttraumatic Sequelae After Acute Medical Events: The Potential of Prognostic Models and Computer-Aided Approaches

Katharina Schultebraucks, PhD; Merlin Bittlinger, MA; Kenrick Cato, PhD; Bernard P. Chang, MD, PhD


The risk of developing adverse mental health effects after acute medical emergencies such as those caused by cardiovascular disease (CVD) and leading to emergency department (ED) admission is an enduring health care challenge with significant public health implications and patient burden. CVD-induced posttraumatic stress disorder is not uncommon and has been found to increase both psychological and cardiac morbidity and mortality. To implement early targeted prevention, a risk-based approach is needed, which requires an accurate clinical prediction model. Risk prediction is best if it relies on a multifactorial model accounting for nonlinear dependencies, which requires advanced computational models. The development of scalable and automated computational prognostic models has the potential to provide clinicians with new tools for accurate and feasible risk stratification directly in the ED. We present recent advancements and discuss the benefits and challenges for the clinical use of algorithm-based risk stratification for posttraumatic aftercare and digital health applications in patients with CVD. [Psychiatr Ann. 2021;51(1):27–32.]


The risk of developing adverse mental health effects after acute medical emergencies such as those caused by cardiovascular disease (CVD) and leading to emergency department (ED) admission is an enduring health care challenge with significant public health implications and patient burden. CVD-induced posttraumatic stress disorder is not uncommon and has been found to increase both psychological and cardiac morbidity and mortality. To implement early targeted prevention, a risk-based approach is needed, which requires an accurate clinical prediction model. Risk prediction is best if it relies on a multifactorial model accounting for nonlinear dependencies, which requires advanced computational models. The development of scalable and automated computational prognostic models has the potential to provide clinicians with new tools for accurate and feasible risk stratification directly in the ED. We present recent advancements and discuss the benefits and challenges for the clinical use of algorithm-based risk stratification for posttraumatic aftercare and digital health applications in patients with CVD. [Psychiatr Ann. 2021;51(1):27–32.]

Cardiovascular disease (CVD) is among the leading causes of mortality and morbidity worldwide, accounting for 1 of 4 deaths in the United States.1 Acute coronary syndrome (ACS) describes a spectrum of CVD that includes acute myocardial infarction or unstable angina.2 In the US, more than 1 million new cases of ACS are diagnosed annually, with many more patients presenting in emergency departments (ED) with suspected ACS.3 Stroke is another important CVD and is the fifth leading cause of death in the US; every year about 800,000 new cases are reported.4 One of the major precursors of stroke is a transient ischemic attack (TIA). There are about 250,000 to 300,000 TIAs reported every year in the US.5

The ED is the first point of contact for many people with acute emergencies, treating about 7.5 million patients with chest pain in 2016.6 The ED is also the frontline for treating patients with stroke and TIA, with 2.5 million ED visits for acute ischemic stroke in the US between 2010 and 20137 and 297,000 annual visits for TIA alone.8 In addition to the well-known physical consequences of CVD, such as medical complications and elevated mortality risk,9 there is also evidence that long-term psychological sequelae can develop after CVD, which, in turn, can increase the risk of subsequent CVD recurrence.10,11 Studies have shown morbidity and mortality of CVD are increased in survivors of CVD who develop posttraumatic stress disorder (PTSD) after the CVD event, with nearly a doubling of all-cause mortality at 1-year follow-up for survivors of ACS who develop PTSD.10–12 In the real-world context of limited and competing resources in the ED, identifying patients with high-risk factors for developing PTSD and preventing PTSD after CVD requires both effective preventive measures as well as accurate prognostic models to identify those patients. This review presents recent findings on risk factors for CVD-induced PTSD and discusses the clinical use of new computational methods for early screening in the ED. These new approaches represent a significant step toward individualized treatment selection using accurate and validated algorithm-based methods for risk-stratification and clinical decision-making. Such a risk-based approach will become key for the proactive and efficient clinical prevention strategy for CVD-related PTSD.

PTSD After CVD Events

Perceived anxiety and stress during a CVD event can lead to the development of psychological stress reactions such as PTSD.10,11 For instance, patients who have a stroke or TIA experience an abrupt and progressive loss of control of the body and cognitive functioning. Patients report having extreme anxiety, shock, and despair during these events.13,14 In contrast to other traumatic events that can result in PTSD, such as disasters, accidents, or sexual assaults, the CVD event is not an external stressor. The unique characteristic of CVD events as traumatic stressors is that these events do not act on the person from the outside but instead the source of the experienced distress is the internal malfunctioning of the body itself, and thus the source is internal to the patients themselves. It has been shown that these internal events such as acute medical incidents can lead to PTSD15 and that patients with CVD at risk of PTSD are also at increased risk of subsequent CVD recurrence.10,11 The enduring somatic threat theory describes the specific characteristics of developing PTSD after acute medical events and emphasizes the psychological importance of the fact that the patient cannot avoid or escape the threat of future events because the source of the traumatic event is somatic and thus internal.16

A meta-analysis of 24 studies in more than 2,300 patients from around the world found that 1 in 8 patients develop ACS-related PTSD.12 Interestingly, patients with suspected ACS but who are later diagnosed as not having ACS have a similar prevalence of PTSD as those who actually experienced ACS.17 A further meta-analysis found that 1 in 4 patients develop stroke-related PTSD within the first year after stroke.18 In patients with cardiac arrest, studies have found that 1 in 3 patients had elevated PTSD symptoms after the event.19 Furthermore, studies have shown that CVD-induced PTSD increases the CVD risk in patients with ACS (hazard ratio 2; 95% confidence interval: 1.7–2.4)12 and in patients with cardiac arrest (hazard ratio 3.2; 95% confidence interval: 1.7–6.0).19 It has also been shown that stroke-induced PTSD is associated with medication nonadherence in a community-sample of 535 patients with TIA or stroke.20 Therefore, there is urgency and high clinical relevance for approaches that would allow early and accurate screening for CVD-induced PTSD in the acute care setting. Recently, new computational methods have emerged to support the prognosis of PTSD risk in the ED, and these methods have a high potential for identifying patients at elevated risk for developing CVD-induced posttraumatic sequelae.

Risk Factors for CVD-Related PTSD

Several risk factors have been discussed for CVD-induced PTSD. Studies have identified demographic risk factors such as younger age, female gender, lower income, and living situation (eg, living alone). Comorbidities such as physical disabilities and medical or psychiatric comorbidities such as prior depression or PTSD have been described as other relevant risk factors. Furthermore, contextual factors have also been identified as risk factors. For instance, exposure to crowding in the ED or perceived severity of stroke has been shown to be an important risk factor for the development of CVD-induced posttraumatic sequelae.21

Patients who perceive the CVD event as life-threatening during the evaluation in the ED have a higher risk of developing posttraumatic sequelae.22 Furthermore, studies have shown that there is a higher PTSD risk in overcrowded EDs,23 with a higher risk in those patients with prior depression.24 Those patients experience crowdedness as more stressful and also report that they have the feeling they received poorer care.24 Moreover, an elevated risk for PTSD has been found in those patients who had the feeling that there was a high likelihood a patient near them in the ED was going to die.25

Importantly, no single factor has demonstrated robust prediction of PTSD course. The development of CVD-induced PTSD is influenced by the combination of multiple risk factors. This is in line with current biopsychosocial models of posttraumatic stress that propose biological, psychological, and social characteristics together contribute to the risk of developing posttraumatic stress sequelae. Computational approaches that are able to take into consideration the influence of these multiple risk factors are necessary. Such a multifactorial model that also accounts for nonlinear dependencies would have a high potential to improve the accuracy of risk stratification.

After-Care Planning for Posttraumatic Sequelae in the ED

Patients in the ED are treated for diverse life-threatening disorders such as stroke and ACS. The ED clinician's work environment is at times stressful and chaotic and marked by cognitively demanding tasks. Given its critical window in the acute evaluation of patients, the ED provides a unique opportunity for early screening for posttraumatic sequelae and a vital time-sensitive window for selective prevention.26 However, traditional screening approaches are time-consuming and, therefore, not feasible in the ED.27 For example, despite the bulk of evidence on the psychological sequelae of acute illness such as PTSD, only 7% of EDs regularly screen for PTSD.28 In the setting of historic demands on EDs, and particularly with additional stressors such as the coronavirus disease 2019 (COVID-19) pandemic, the prospective screening for acute stress symptoms encounters challenges with both a clinician's working capacities and time resources, making the implementation of such clinician-executed screening protocols difficult. In contrast, computational models using routinely collectible data would not face this challenge and would provide clinicians with prognostic models that also would work in the context of natural disasters or crises.

Furthermore, mounting evidence shows that only a minority of people develop CVD-induced PTSD.12,18 Individualized risk stratification and treatment selection are thus needed, as blanket interventions are not indicated and not cost-effective.26 Computerized prognostic models can help to accurately define and scale risk stratification and help to prioritize interventions. Those feasible and scalable prognostic models for posttraumatic sequelae after medical emergencies are urgently needed, as the current COVID-19 pandemic shows.

Algorithm-Based Risk Stratification

New and innovative approaches are needed to develop accurate prognostic models, especially the ability to harness clinical data already collected in the ED to reduce the additional burden for clinicians. Machine learning and artificial intelligence (AI) have been shown to be beneficial for health care in many medical domains.

Studies have also shown that data collected in the ED can be used for risk stratification. For instance, studies have used natural language processing on electronic health records (EHRs)29,30 as a promising method to collect clinically informative data without posing burdens to the standards of operating procedures of the ED.

The development of an automated algorithm that can be integrated into routine clinical care would be of high importance. A recent study in survivors of trauma admitted to the ED showed the feasibility of predicting long-term PTSD and also showed that the prediction accuracy was generalizable to an independent external sample of trauma survivors.31 The researchers also explored the algorithm's predictive value, using routinely collected data from EHRs in the ED setting,31 which has the potential to be implemented as a rapid and automated clinical readout available for ED clinicians and those responsible for discharge planning.

The use of clinical decision support (CDS) tools that deploy algorithm-based prediction using routinely collectible data has opened promising opportunities to provide ED clinicians with a timely prognostic model of adverse mental health effects that will allow clinicians to improve the planning of after-care and preventive measures. These AI-based CDS tools will provide clinical decision support in the form of EHR-embedded apps or alerts. For example, the recent research in the area of algorithm-based risk prediction in the ED is promising foundational work for the development of CDS tools that will someday indicate to ED clinicians which patients are at higher risk for PTSD, allowing for the appropriate follow-up care.31

Individualized Interventions and Digital Health Approaches

Well-established interventions exist for PTSD. Cognitive-behavioral therapy is effective, with an estimated number needed-to-treat of 5, but this type of therapy is also costly.32 The same is true for early preventive interventions in trauma survivors,26 such as early exposure intervention in the ED shortly after the trauma happened, as well as early internet-based intervention and pharmacological interventions.26 Whether such interventions are also effective for special populations, such as patients with CVD-induced PTSD, and which therapeutic approach is best suited, remains to be tested.33

Digital health approaches are of high relevance because they offer the opportunity to remotely monitor and classify psychopathology and, therefore, remove barriers and increase scalability.34 The implementation of digital health approaches after CVD events has the potential to monitor and treat CVD-induced psychological sequelae and to reduce the risk of future morbidity and mortality of patients at high risk. This is of high importance because CVD-induced PTSD symptoms are related to an aversion to medications, as reflected in aversive cognitions.35 Furthermore, CVD-induced PTSD is related to an aversion to physical activity (adjusted odds ratio, 1.4–2.0; P < .01 in those patients with PTSD vs those without PTSD).36 These factors in turn influence the risk of future CVD events. There is a high need for evidence-based digital health applications after acute medical emergencies to influence the trajectory of posttraumatic sequelae.37

Telehealth applications for CVD patients are broad and timely. Within the past 10 years, the percentage of US adults owning a smartphone has increased from 35% to more than 80%,3 creating a digital environment allowing for rapid transmission of information and continuous communication.38 This digital environment also creates a unique opportunity for rapid data-driven testing and dissemination of effective solutions to assess and treat the mental health problems anticipated to arise during the wake of any acute medical events. Digital mental health programs offer the ability to respond quickly and efficiently and to reach people over great distances with minimal mobility requirements.39 The range of digital approaches to health is broad. These include guided telemedicine encounters with a licensed clinician (eg, therapy/treatments conducted over video or phone), computer-guided internet/app-based mental health programs with nonclinician coaches, and finally unguided internet/app-based therapies with no coach or therapist.5 For survivors of CVD, early evidence has already found that digital health approaches have been associated with improvements in CVD risk factors such as blood pressure40 and weight loss.41

Challenges and Benefits of Automated Risk Prediction

Using prognostic models in an acute care setting provides several benefits to clinicians and patients. The use of algorithm-based risk stratification is also of ethical importance and poses several challenges that need to be considered.

An algorithm is only as good as the data it relies on. This is particularly important when the data stems from EHRs and other data sources that are not specifically collected for the purpose of risk prediction but are nonetheless used for this purpose. Because data quality plays such a central role, it is of utmost importance that quality checks are applied at every step of algorithm development, testing, and application. Such quality checks are necessary to prevent, detect, and then eliminate potential biases in the algorithm. For example, studies have shown racial and algorithmic biases in commercial prediction models.42 Because risk prediction models are based on observational data, the risk for such biases can arise from potential confounding factors that cannot be controlled for in observation studies. It is, therefore, essential that the data used for the development of the algorithm cover a diverse and heterogeneous population that is as representative as possible of the target population to which the algorithm will be applied. In addition, it is important to always critically appraise on what patient population the algorithm was initially developed before the algorithm is applied to individual patients or cohorts of patients that are known to significantly deviate from the patient population of the original model. So, clinicians who apply such an algorithm must be informed about the characteristics of the model development patient population and receive guidance for which patient population the algorithm is best suited. Moreover, the performance of any algorithm should be frequently reevaluated to spot systematic factors that may introduce bias into the predictions, such as systematic underrepresentation of racial and ethnic minority populations. With such safeguards in place, an accurate prediction algorithm has great potential to significantly improve risk prediction and can be an important foundation for CDS tools with high clinical utility.


There is a significant risk for the development of posttraumatic sequelae after CVD that leads to a higher risk of future morbidity and mortality in these patients. Algorithm-based risk stratification in the acute care setting and the use of digital health applications have the promise to provide clinical decision support to improve the follow-up care of these patients with high-risk factors and, therefore, have the potential to decrease disease burden and mortality.


  1. Centers for Disease Control and Prevention. Multiple causes of death 1999–2018. Accessed December 9, 2020.
  2. Fanaroff AC, Rymer JA, Goldstein SA, Simel DL, Newby LK. Does this patient with chest pain have acute coronary syndrome?: the rational clinical examination systematic review. JAMA. 2015;314(18):1955–1965. doi:10.1001/jama.2015.12735 [CrossRef] PMID:26547467
  3. Benjamin EJ, Muntner P, Alonso A, et al. American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics-2019 update: a report from the American Heart Association. Circulation. 2019;139(10):e56–e528. doi:10.1161/CIR.0000000000000659 [CrossRef] PMID:30700139
  4. Mozaffarian D, Benjamin EJ, Go AS, et al. Writing Group MembersAmerican Heart Association Statistics CommitteeStroke Statistics Subcommittee. Executive summary: heart disease and stroke statistics—2016 update: a report from the American Heart Association. Circulation. 2016;133(4):447–454. doi:10.1161/CIR.0000000000000366 [CrossRef] PMID:26811276
  5. Powers WJ, Rabinstein AA, Ackerson T, et al. 2018 Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2018;49(3):e46–e110. doi:10.1161/STR.0000000000000158 [CrossRef] PMID:29367334
  6. Rui P, Kang K, Ashman J. National hospital ambulatory medical care survey: 2016 emergency department summary tables. Accessed December 5, 2020.
  7. Stuntz M, Busko K, Irshad S, Paige T, Razhkova V, Coan T. Nationwide trends of clinical characteristics and economic burden of emergency department visits due to acute ischemic stroke. Open Access Emerg Med. 2017;9:89–96. doi:10.2147/OAEM.S146654 [CrossRef] PMID:29033616
  8. Edlow JA, Kim S, Pelletier AJ, Camargo CA Jr, . National study on emergency department visits for transient ischemic attack, 1992–2001. Acad Emerg Med. 2006;13(6):666–672. PMID:16609106
  9. Dennis MS, Burn JP, Sandercock PA, Bamford JM, Wade DT, Warlow CP. Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1993;24(6):796–800. doi:10.1161/01.STR.24.6.796 [CrossRef] PMID:8506550
  10. Schultebraucks K, Wen T, Kronish IM, Willey J, Chang BP. Post-traumatic stress disorder following acute stroke. Curr Emerg Hosp Med Rep. 2020;8(1):1–8. doi:10.1007/s40138-020-00206-5 [CrossRef]
  11. Musey PI Jr, Schultebraucks K, Chang BP. Stressing out about the heart: a narrative review of the role of psychological stress in acute cardiovascular events. Acad Emerg Med. 2020;27(1):71–79. doi:10.1111/acem.13882 [CrossRef] PMID:31675448
  12. Edmondson D, Richardson S, Falzon L, Davidson KW, Mills MA, Neria Y. Post-traumatic stress disorder prevalence and risk of recurrence in acute coronary syndrome patients: a meta-analytic review. PLoS One. 2012;7(6):e38915. doi:10.1371/journal.pone.0038915 [CrossRef] PMID:22745687
  13. Goldfinger JZ, Edmondson D, Kronish IM, et al. Correlates of post-traumatic stress disorder in stroke survivors. J Stroke Cerebrovasc Dis. 2014;23(5):1099–1105. doi:10.1016/j.jstrokecerebrovasdis.2013.09.019 [CrossRef] PMID:24144593
  14. Chang BP. Can hospitalization be hazardous to your health? A nosocomial based stress model for hospitalization. Gen Hosp Psychiatry. 2019;60:83–89. doi:10.1016/j.genhosppsych.2019.07.014 [CrossRef] PMID:31376645
  15. Shalev AY, Schreiber S, Galai T, Melmed RN. Post-traumatic stress disorder following medical events. Br J Clin Psychol. 1993;32(2):247–253. doi:10.1111/j.2044-8260.1993.tb01052.x [CrossRef] PMID:8318945
  16. Edmondson D. An enduring somatic threat model of posttraumatic stress disorder due to acute life threatening medical events. Soc Personal Psychol Compass. 2014;8(3):118–134. doi:10.1111/spc3.12089 [CrossRef] PMID:24920956
  17. Kronish IM, Edmondson D, Moise N, et al. Posttraumatic stress disorder in patients who rule out versus rule in for acute coronary syndrome. Gen Hosp Psychiatry. 2018;53:101–107. doi:10.1016/j.genhosp-psych.2018.02.007 [CrossRef] PMID:29773309
  18. Edmondson D, Richardson S, Fausett JK, Falzon L, Howard VJ, Kronish IM. Prevalence of PTSD in survivors of stroke and transient ischemic attack: a meta-analytic review. PLoS One. 2013;8(6):e66435. doi:10.1371/journal.pone.0066435 [CrossRef] PMID:23840467
  19. Agarwal S, Presciutti A, Cornelius T, et al. Cardiac arrest and subsequent hospitalization-induced posttraumatic stress is associated with 1-year risk of major adverse cardiovascular events and all-cause mortality. Crit Care Med. 2019;47(6):e502–e505. doi:10.1097/CCM.0000000000003713 [CrossRef] PMID:30889030
  20. Kronish IM, Edmondson D, Goldfinger JZ, Fei K, Horowitz CR. Posttraumatic stress disorder and adherence to medications in survivors of strokes and transient ischemic attacks. Stroke. 2012;43(8):2192–2197. doi:10.1161/STROKEAHA.112.655209 [CrossRef] PMID:22618380
  21. Edmondson D, von Känel R. Post-traumatic stress disorder and cardiovascular disease. Lancet Psychiatry. 2017;4(4):320–329. doi:10.1016/S2215-0366(16)30377-7 [CrossRef] PMID:28109646
  22. Zhu DR, Julian J, Lee SJA, Thanataveerat A, Sumner JA. Patterns of peritraumatic threat perceptions in patients evaluated for suspected acute coronary syndrome according to prior and current posttraumatic stress symptoms. Gen Hosp Psychiatry. 2018;53:119–124. doi:10.1016/j.genhosp-psych.2018.03.003 [CrossRef] PMID:29789141
  23. Edmondson D, Shimbo D, Ye S, Wyer P, Davidson KW. The association of emergency department crowding during treatment for acute coronary syndrome with subsequent post-traumatic stress disorder symptoms. JAMA Intern Med. 2013;173(6):472–474. doi:10.1001/jamainternmed.2013.2536 [CrossRef] PMID:23400256
  24. St Onge T, Edmondson D, Cea E, Husain S, Chang BP. Depressive symptoms and perceptions of ED care in patients evaluated for acute coronary syndrome. J Emerg Nurs. 2018;44(1):46–51. doi:10.1016/j.jen.2017.04.010 [CrossRef] PMID:28527642
  25. Konrad B, Hiti D, Chang BP, Retuerto J, Julian J, Edmondson D. Cardiac patients' perceptions of neighboring patients' risk: influence on psychological stress in the ED and subsequent posttraumatic stress. BMC Emerg Med. 2017;17(1):33. doi:10.1186/s12873-017-0144-3 [CrossRef] PMID:29110718
  26. Shalev AY, Barbano AC. PTSD: risk assessment and early management. Psychiatr Ann. 2019;49(7):299–306. doi:10.3928/00485713-20190605-01 [CrossRef]
  27. van der Mei WF, Barbano AC, Ratanatharathorn A, et al. International Consortium to Predict PTSD. Evaluating a screener to quantify PTSD risk using emergency care information: a proof of concept study. BMC Emerg Med. 2020;20(1):16. doi:10.1186/s12873-020-00308-z [CrossRef] PMID:32122334
  28. Love J, Zatzick D. Screening and intervention for comorbid substance disorders, PTSD, depression, and suicide: a trauma center survey. Psychiatr Serv. 2014;65(7):918–923. doi:10.1176/ [CrossRef] PMID:24733143
  29. Korach ZT, Yang J, Rossetti SC, et al. Mining clinical phrases from nursing notes to discover risk factors of patient deterioration. Int J Med Inform. 2020;135:104053. doi:10.1016/j.ijmedinf.2019.104053 [CrossRef] PMID:31884312
  30. Topaz M, Murga L, Bar-Bachar O, Cato K, Collins S. Extracting alcohol and substance abuse status from clinical notes: the added value of nursing data. Stud Health Technol Inform. 2019;264:1056–1060. doi:10.3233/SHTI190386 [CrossRef] PMID:31438086
  31. Schultebraucks K, Shalev AY, Michopoulos V, et al. A validated predictive algorithm of post-traumatic stress course following emergency department admission after a traumatic stressor. Nat Med. 2020;26(7):1084–1088. doi:10.1038/s41591-020-0951-z [CrossRef] PMID:32632194
  32. Forman-Hoffman V, Cook Middleton J, Feltner C, et al. Psychological and pharmacological treatments for adults with posttraumatic stress disorder: a systematic review update. Agency for Healthcare Research and Quality; 2018. doi:10.23970/AHRQEPCCER207 [CrossRef]
  33. Birk JL, Sumner JA, Haerizadeh M, et al. Early interventions to prevent posttraumatic stress disorder symptoms in survivors of life-threatening medical events: a systematic review. J Anxiety Disord. 2019;64:24–39. doi:10.1016/j.janxdis.2019.03.003 [CrossRef] PMID:30925334
  34. Carmi L, Schultebraucks K, Galatzer-Levy I. Identification, prediction, and intervention via remote digital technology: digital phenotyping & deployment of clinical interventions following terror and mass casualty events. In: Vermetten E, Frankova I, Carmi L, Chaban O, Zohar J, eds. Management of Terrorism Induced Stress – Guideline for the Golden Hours. Vol 148. IOS Press BV; 2020:175–181.
  35. Husain SA, Edmondson D, Kautz M, Umland R, Kronish IM. Posttraumatic stress disorder due to acute cardiac events and aversive cognitions towards cardiovascular medications. J Behav Med. 2018;41(2):261–268. doi:10.1007/s10865-017-9906-3 [CrossRef] PMID:29204908
  36. Monane R, Sanchez GJ, Kronish IM, Edmondson D, Diaz KM. Post-traumatic stress disorder symptoms and aversive cognitions regarding physical activity in patients evaluated for acute coronary syndrome. Eur J Prev Cardiol. 2018;25(4):402–403. doi:10.1177/2047487317746255 [CrossRef] PMID:29198138
  37. Chang BP, Kessler RC, Pincus HA, Nock MK. Digital approaches for mental health in the age of Covid-19. BMJ. 2020;369:m2541. doi:10.1136/bmj.m2541 [CrossRef] PMID:32601049
  38. Liu F, Zhao S, Li Y. How many, how often, and how new? A multivariate profiling of mobile app users. J Retailing Consum Serv. 2017;38:71–80. doi:10.1016/j.jretconser.2017.05.008 [CrossRef]
  39. Marzano L, Bardill A, Fields B, et al. The application of mHealth to mental health: opportunities and challenges. Lancet Psychiatry. 2015;2(10):942–948. doi:10.1016/S2215-0366(15)00268-0 [CrossRef] PMID:26462228
  40. Harzand A, Witbrodt B, Davis-Watts ML, et al. Feasibility of a smartphone-enabled cardiac rehabilitation program in male veterans with previous clinical evidence of coronary heart disease. Am J Cardiol. 2018;122(9):1471–1476. doi:10.1016/j.amjcard.2018.07.028 [CrossRef] PMID:30217377
  41. Toro-Ramos T, Lee D-H, Kim Y, et al. Effectiveness of a smartphone application for the management of metabolic syndrome components focusing on weight loss: a preliminary study. Metab Syndr Relat Disord. 2017;15(9):465–473. doi:10.1089/met.2017.0062 [CrossRef] PMID:29035677
  42. Obermeyer Z, Powers B, Vogeli C, Mullainathan S. Dissecting racial bias in an algorithm used to manage the health of populations. Science. 2019;366(6464):447–453. doi:10.1126/science.aax2342 [CrossRef] PMID:31649194

Katharina Schultebraucks, PhD, is the Director of Computational Medicine and Artificial Intelligence and an Assistant Professor in Behavioral and Cognitive Sciences, Department of Emergency Medicine, Columbia University Medical Center. Merlin Bittlinger, MA, is a Research Associate, Charité – Universitätsmedizin. Kenrick Cato, PhD, is an Assistant Professor, School of Nursing, Columbia University Irving Medical Center. Bernard P. Chang, MD, PhD, is an Associate Professor, Department of Emergency Medicine, Columbia University.

Address correspondence to Katharina Schultebraucks, PhD, Department of Emergency Medicine, Columbia University Medical Center, 622 West 168th Street, New York, NY 10032, email:

Disclosure: The authors have no relevant financial relationships to disclose.


Sign up to receive

Journal E-contents