Clinical Decision Support Software Research Articles

Strengthening the use of artificial intelligence within healthcare delivery organizations: balancing regulatory compliance and patient safety.

Surface the urgent dilemma that healthcare delivery organizations (HDOs) face navigating the US Food and Drug Administration (FDA) final guidance on the use of clinical decision support (CDS) software. We use sepsis as a case study to highlight the patient safety and regulatory compliance tradeoffs that 6129 hospitals in the United States must navigate. Sepsis CDS remains in broad, routine use. There is no commercially available sepsis CDS system that is FDA cleared as a medical device. There is no public disclosure of an HDO turning off sepsis CDS due to regulatory compliance concerns. And there is no public disclosure of FDA enforcement action against an HDO for using sepsis CDS that is not cleared as a medical device. We present multiple policy interventions that would relieve the current tension to enable HDOs to utilize artificial intelligence to improve patient care while also addressing FDA concerns about product safety, efficacy, and equity.

An Assessment of Different Decision Support Software from the Perspective of Potential Drug-Drug Interactions in Patients with Chronic Kidney Diseases.

Chronic kidney disease (CKD) is a multifaceted disorder influenced by various factors. Drug-drug interactions (DDIs) present a notable risk factor for hospitalization among patients with CKD. This study aimed to assess the frequency and attributes of potential DDIs (pDDIs) in patients with CKD and to ascertain the concordance among different Clinical Decision Support Software (CDSS). A cross-sectional study was conducted in a nephrology outpatient clinic at a university hospital. The pDDIs were identified and evaluated using Lexicomp® and Medscape®. The patients' characteristics, comorbidities, and medicines used were recorded. The concordance of different CDSS were evaluated using the Kendall W coefficient. An evaluation of 1121 prescribed medications for 137 patients was carried out. The mean age of the patients was 64.80 ± 14.59 years, and 41.60% of them were male. The average year with CKD was 6.48 ± 5.66. The mean number of comorbidities was 2.28 ± 1.14. The most common comorbidities were hypertension, diabetes, and coronary artery disease. According to Medscape, 679 pDDIs were identified; 1 of them was contraindicated (0.14%), 28 (4.12%) were serious-use alternative, and 650 (9.72%) were interventions that required closely monitoring. According to Lexicomp, there were 604 drug-drug interactions. Of these interactions, 9 (1.49%) were in the X category, 60 (9.93%) were in the D category, and 535 (88.57%) were in the C category. Two different CDSS systems exhibited statistically significant concordance with poor agreement (W = 0.073, p < 0.001). Different CDSS systems are commonly used in clinical practice to detect pDDIs. However, various factors such as the operating principles of these programs and patient characteristics can lead to incorrect guidance in clinical decision making. Therefore, instead of solely relying on programs with lower reliability and consistency scores, multidisciplinary healthcare teams, including clinical pharmacists, should take an active role in identifying and preventing pDDIs.

Medication Deprescribing in Patients Receiving Hemodialysis: A Prospective Controlled Quality Improvement Study

Patients treated with dialysis are commonly prescribed multiple medications (polypharmacy), including some potentially inappropriate medications (PIMs). PIMs are associated with an increased risk of medication harm (eg, falls, fractures, hospitalization). Deprescribing is a solution that proposes to stop, reduce, or switch medications to a safer alternative. Although deprescribing pairs well with routine medication reviews, it can be complex and time-consuming. Whether clinical decision support improves the process and increases deprescribing for patients treated with dialysis is unknown. This study aimed to test the efficacy of the clinical decision support software MedSafer at increasing deprescribing for patients treated with dialysis. Prospective controlled quality improvement study with a contemporaneous control. Patients prescribed≥5 medications in 2 outpatient dialysis units in Montréal, Canada. Patient health data from the electronic medical record were input into the MedSafer web-based portal to generate reports listing candidate PIMs for deprescribing. At the time of a planned biannual medication review (usual care), treating nephrologists in the intervention unit additionally received deprescribing reports, and patients received EMPOWER brochures containing safety information on PIMs they were prescribed. In the control unit, patients received usual care alone. The proportion of patients with≥1 PIMs deprescribed was compared between the intervention and control units following a planned medication review to determine the effect of using MedSafer. The absolute risk difference with 95% CI and number needed to treat were calculated. The primary outcome was the proportion of patients with one or more PIMs deprescribed. Secondary outcomes include the reduction in the mean number of prescribed drugs and PIMs from baseline. In total, 195 patients were included (127, control unit; 68, intervention unit); the mean age was 64.8±15.9 (SD), and 36.9% were women. The proportion of patients with≥1 PIMs deprescribed in the control unit was 3.1% (4/127) vs 39.7% (27/68) in the intervention unit (absolute risk difference, 36.6%; 95% CI, 24.5%-48.6%; P<0.0001; number needed to treat=3). This was a single-center nonrandomized study with a type 1 error risk. Deprescribing durability was not assessed, and the study was not powered to reduce adverse drug events. Deprescribing clinical decision support and patient EMPOWER brochures provided during medication reviews could be an effective and scalable intervention to address PIMs in the dialysis population. A confirmatory randomized controlled trial is needed.

Optimising diagnosis and management of kidney disease: an implementation trial of a clinical decision support system future health today

BackgroundChronic kidney disease affects more than 10% of the world’s population and is a non-communicable disease of global concern and priority. There is a significant implementation gap between best practice guideline recommendations and current kidney disease management. Previous research has shown the need to partner with primary care to improve education, collaboration, and kidney disease awareness. This implementation trial will explore use of an innovative clinical decision support software, Future Health Today, to improve screening, diagnosis, and management of kidney disease in primary care. The program will be supported by tertiary care outreach services. The primary aim is to test the hypothesis that the Future Health Today implementation program will improve screening, diagnosis, and management of kidney disease. Secondary aims are to evaluate primary care satisfaction and broader health service impacts.MethodsThis pre-post implementation trial using an interrupted time series design will evaluate the clinical and service outcomes of Future Health Today, using a mixed methods study in twenty general practices with an estimated population size of 150,000. Deidentified patient data will be extracted from participating practices to examine the primary aims of the study. Surveys and semi-structured interviews with general practice will inform secondary hypotheses. Data linkage between primary care and tertiary care data will examine the broader health service impacts.DiscussionThis investigator driven trial will assess the impact of Future Health Today software coupled with education and clinical outreach support. Investigators hypothesise that there will be improvement in appropriate screening, diagnosis, and management of kidney disease. This program has the potential to be scaled more broadly.Trial RegistrationAustralian New Zealand Clinical Trial Registry: ACTRN12623001096640.

Evaluating clinical decision support software (CDSS): challenges for robust evidence generation.

Computerized clinical decision support software (CDSS) are digital health technologies that have been traditionally categorized as medical devices. However, the evaluation frameworks for traditional medical devices are not well adapted to assess the value and safety of CDSS. In this study, we identified a range of challenges associated with CDSS evaluation as a medical device and investigated whether and how CDSS are evaluated in Australia. Using a qualitative approach, we interviewed 11 professionals involved in the implementation and evaluation of digital health technologies at national and regional levels. Data were thematically analyzed using both data-driven (inductive) and theory-based (deductive) approaches. Our results suggest that current CDSS evaluations have an overly narrow perspective on the risks and benefits of CDSS due to an inability to capture the impact of the technology on the sociotechnical environment. By adopting a static view of the CDSS, these evaluation frameworks are unable to discern how rapidly evolving technologies and a dynamic clinical environment can impact CDSS performance. After software upgrades, CDSS can transition from providing information to specifying diagnoses and treatments. Therefore, it is not clear how CDSS can be monitored continuously when changes in the software can directly affect patient safety. Our findings emphasize the importance of taking a living health technology assessment approach to the evaluation of digital health technologies that evolve rapidly. There is a role for observational (real-world) evidence to understand the impact of changes to the technology and the sociotechnical environment on CDSS performance.

Application of a human-centered design for embedded machine learning model to develop data labeling software with nurses: Human-to-Artificial Intelligence (H2AI)

BackgroundNurses are essential for assessing and managing acute pain in hospitalized patients, especially those who are unable to self-report pain. Given their role and subject matter expertise (SME), nurses are also essential for the design and development of a supervised machine learning (ML) model for pain detection and clinical decision support software (CDSS) in a pain recognition automated monitoring system (PRAMS). Our first step for developing PRAMS with nurses was to create SME-friendly data labeling software. PurposeTo develop an intuitive and efficient data labeling software solution, Human-to-Artificial Intelligence (H2AI). MethodThe Human-centered Design for Embedded Machine Learning Solutions (HCDe-MLS) model was used to engage nurses. In this paper, HCDe-MLS will be explained using H2AI and PRAMS as illustrative cases. FindingsUsing HCDe-MLS, H2AI was developed and facilitated labeling of 139 videos (mean = 29.83 min) with 3189 images labeled (mean = 75 s) by 6 nurses. OpenCV was used for video-to-image pre-processing; and MobileFaceNet was used for default landmark placement on images. H2AI randomly assigned videos to nurses for data labeling, tracked labelers’ inter-rater reliability, and stored labeled data to train ML models. ConclusionsNurses’ engagement in CDSS development was critical for ensuring the end-product addressed nurses’ priorities, reflected nurses’ cognitive and decision-making processes, and garnered nurses’ trust for technology adoption.

Semi-automation of keratopathy visual acuity grading of corneal events in belantamab mafodotin clinical trials: clinical decision support software.

Belantamab mafodotin (belamaf) has demonstrated clinically meaningful antimyeloma activity in patients with heavily pretreated multiple myeloma. However, it is highly active against dividing cells, contributing to off-target adverse events, particularly ocular toxicity. Changes in best corrected visual acuity (BCVA) and corneal examination findings are routinely monitored to determine Keratopathy Visual Acuity (KVA) grade to inform belamaf dose modification. We aimed to develop a semiautomated mobile app to facilitate the grading of ocular events in clinical trials involving belamaf. The paper process was semiautomated by creating a library of finite-state automaton (FSA) models to represent all permutations of KVA grade changes from baseline BCVA readings. The transition states in the FSA models operated independently of eye measurement units (e.g., Snellen, logMAR, decimal) and provided a uniform approach to determining KVA grade changes. Together with the FSA, the complex decision tree for determining the grade change based on corneal examination findings was converted into logical statements for accurate and efficient overall KVA grade computation. First, a web-based user interface, conforming to clinical practice settings, was developed to simplify the input of key KVA grading criteria. Subsequently, a mobile app was developed that included additional guided steps to assist in clinical decision-making. The app underwent a robust Good Clinical Practice validation process. Outcomes were reviewed by key stakeholders, our belamaf medical lead, and the systems integration team. The time to compute a patient's overall KVA grade using the Belamaf Eye Exam (BEE) app was reduced from a 20- to 30-min process to <1-2 min. The BEE app was well received, with most investigators surveyed selecting "satisfied" or "highly satisfied" for its accuracy and time efficiency. Our semiautomated approach provides for an accurate, simplified method of assessment of patients' corneal status that reduces errors and quickly delivers information critical for potential belamaf dose modifications. The app is currently available on the Apple iOS and Android platforms for use by investigators of the DREAMM clinical trials, and its use could easily be extended to the clinic to support healthcare providers who need to make informed belamaf treatment decisions.

Modelling clinical narrative as computable knowledge: The NICE computable implementation guidance project.

Translating narrative clinical guidelines to computable knowledge is a long-standing challenge that has seen a diverse range of approaches. The UK National Institute for Health and Care Excellence (NICE) Content Advisory Board (CAB) aims ultimately to (1) guide clinical decision support and other software developers to increase traceability, fidelity and consistency in supporting clinical use of NICE recommendations, (2) guide local practice audit and intervention to reduce unwarranted variation, (3) provide feedback to NICE on how future recommendations should be developed. The first phase of work was to explore a range of technical approaches to transition NICE toward the production of natively digital content. Following an initial 'collaborathon' in November 2022, the NICE Computable Implementation Guidance project (NCIG) was established. We held a series of workstream calls approximately fortnightly, focusing on (1) user stories and trigger events, (2) information model and definitions, (3) horizon-scanning and output format. A second collaborathon was held in March 2023 to consolidate progress across the workstreams and agree residual actions to complete. While we initially focussed on technical implementation standards, we decided that an intermediate logical model was a more achievable first step in the journey from narrative to fully computable representation. NCIG adopted the WHO Digital Adaptation Kit (DAK) as a technology-agnostic method to model user scenarios, personae, processes and workflow, core data elements and decision-support logic. Further work will address indicators, such as prescribing compliance, and implementation in document templates for primary care patient record systems. The project has shown that the WHO DAK, with some modification, is a promising approach to build technology-neutral logical specifications of NICE recommendations. Implementation of concurrent computable modelling by multidisciplinary teams during guideline development poses methodological and cultural questions that are complex but tractable given suitable will and leadership.

A clinical decision support system to predict the efficacy for EGFR-TKIs based on artificial neural network.

The efficacy of epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitor (TKI) was affected by numerous factors. In the study, we developed and validated an artificial neural network (ANN) system based on clinical characteristics and next-generation sequencing (NGS) to support clinical decisions. A multicenter retrospective non-interventional study was conducted. 240 patients from three hospitals with advanced non-small cell lung cancer (NSCLC) and EGFR mutation were tested by NGS before the first treatment. All patients received formal EGFR-TKIs treatment. Five different models were individually trained to predict the efficacy of EGFR-TKIs based on one medical center with 188 patients. Two independent cohorts from other medical centers were collected for external validation. Compared with logistic regression, four machine learning methods showed better predicting abilities for EGFR-TKIs. The inclusion of NGS tests improved the predictive power of models. ANN performed best on the dataset with mutations TP53, RB1, PIK3CA, EGFR mutation sites, and tumor mutation burden (TMB). The prediction accuracy, recall and AUC were 0.82, 0.82, and 0.82, respectively in our final model. In the external validation set, ANN still showed good performance and differentiated patients with poor outcomes. Finally, a clinical decision support software based on ANN was developed and provided a visualization interface for clinicians. This study provides an approach to assess the efficacy of NSCLC patients with first-line EGFR-TKI treatment. Software is developed to support clinical decisions.

Impact of the Regulatory Framework on Medical Device Software Manufacturers: Are the Guidance Documents Supporting the Practical Implementation? Comment on "Clinical Decision Support and New Regulatory Frameworks for Medical Devices: Are We Ready for It? - A Viewpoint Paper".

The increasing use in clinical practice of software such as mobile apps and clinical decision support (CDS) software has only recently been taken up by regulators around the world. Specifically, the European Commission and the US Food and Drug Administration (FDA) have updated their regulatory framework in the last years. Van Laere et al have given an extensive overview of the European and US approaches to regulate CDS software. This commentary further discusses regulatory differences between the two geographies and their impact on manufacturers of medical device software. We discuss the practical implementation of the regulatory framework for medical device software (especially CDS software) with a reference to the available international guidance documents and their limitations. Given the direction of stricter regulatory oversight in Europe, additional European guidelines/examples are desirable to enable a pragmatic regulatory approach ensuring continued access to innovative medical device software for European patients.

What effects did home working have on 999 clinician practice from one UK ambulance service during the Covid-19 pandemic?

PurposeVirtual care is any interaction between a patient and clinician or clinicians, occurring remotely using information technologies. Like many international services, United Kingdom (UK) ambulance services use paramedics and nurses to undertake telephone assessments of patients calling the 999 emergency service line. Using their clinical knowledge, experience, and, at times, computer decision support software, these clinicians assess patients from a range of clinical acuities to confirm the need for an emergency response or identify and support those patients who can be cared for with remote treatment advice and referral. The Covid-19 pandemic saw UK ambulance services change and adapt their operating models to meet social distancing requirements, increase clinical staff numbers and mitigate staff becoming unavailable for work due to self-isolation. One such strategy was moving clinicians from Emergency Operations Centres (EOCs) to working at home. Staff utilised digital phone systems, remote computer-aided dispatch modules, remote clinical decision support software and video platforms, which allowed close to full functionality compared to inside typical EOCs. There is a dearth of literature exploring the comparative practice of clinicians from home rather than from EOCs. Therefore, this study reports the findings of a qualitative analysis of these effects from the clinician's perspective. The authors hope that the findings from this study will inform the operating, education and leadership practices of those delivering such services.Design/methodology/approachA convenience sample of telephone nurses and paramedics from one UK ambulance service in which home working had been implemented were contacted. 15 clinicians with recent home working experience responded to the invite out of a possible 31 (48%). All participants had previously practised remote assessment from within an EOC. Semi-structured interviews took place via video conferencing software and were recorded, transcribed and thematically analysed. An inductive approach was taken to generating codes and both researchers separately reading the transcripts before re-reading them, assigning initial themes and determining frequency.FindingsFour main themes were identified with further associated sub-themes: (1) performance, (2) support, (3) distractions and interruptions and (4) confidence in decision-making.Originality/valueThere are very few studies exploring the practice of remote clinicians in emergency EOCs. This study identified that home working clinicians felt their productivity had increased, making them more satisfied in their practice. However, there were mixed feelings over the level of support they perceived they now received, despite the mechanisms of support being largely the same. Supervisors found it especially challenging to provide support to practitioners; and employers might need to clarify the support mechanisms they provide to homeworkers. The elimination of distractions and interruptions was seen as a largely positive result of homeworking; however, these interruptions were not seen as inappropriate, thus, identifying a need for role clarity and task coordination rather than interruption elimination. Finally, clinicians felt that they become more confident when working from home, researching more, trusting themselves more and relying less on others to reach safe outcomes. However, there were missed opportunities to learn from listening to others' clinical practice.

CO221 Efficacy and Safety of Personalized Topography-Guided-Lasik Laser Refractive Correction: A Targeted Literature Review

Topography-guided Laser Assisted In-situ Keratomileusis (TG-LASIK) provides patients with corrected visual acuity and quality within a short period after surgery. A targeted literature review assessed the efficacy and safety of personalized TG-LASIK and its use with clinical decision support software.

EE85 A Cost-Utility Analysis Comparing a Clinical Decision Support Software to Prescriber Decision Alone for Drug-Drug Interactions from an U.S. Payer Perspective

To avoid the potential harm of drug-drug interactions (DDIs), clinical decision support (CDS) systems using electronic health records have been installed in multiple healthcare systems. We aim to estimate whether DDI-CDS systems are cost-effective compared to prescribers’ decisions alone considering DDIs.

Online Clinical Decision Support Software Reduces Time to Diagnosis of Fertility Disorders to 8 Weeks on Average [ID: 1380156

INTRODUCTION: Time to diagnosis of fertility disorders is reported to take up to 11 years and involves five to six physicians on average. Prolonged time to diagnosis is not just an immense burden on the health care system, but also means that in vitro fertilization (IVF) becomes necessary for couples who would have been able to conceive naturally had they been diagnosed sooner, often leaving these couples childless. This observational study evaluated how a clinical decision support software based on American Society for Reproductive Medicine, American College of Obstetricians and Gynecologists, and European Society of Human Reproduction and Embryology guidelines used by patients and obstetrician–gynecologists together can reduce the time to diagnosis of up to 65 conditions affecting female fertility. METHODS: One hundred eighty women aged 20–45 who had been trying to conceive for an average of 11 months were recruited to participate. Women completed a reproductive health assessment with a dynamic online questionnaire. The output of the assessment was a lab requisition with an individualized selection of female and male sex hormones, thyroid hormones, markers for glucose metabolism, vitamin deficiencies, complete blood count, as well as markers for immunology, genetics, and thrombophilia. The clinical decision support software combined the results from the online reproductive health assessment with the lab results to render recommendations for diagnosis, further diagnostics, and treatment. RESULTS: An average of four previously unknown health conditions affecting fertility were diagnosed with an average time of 8 weeks to diagnosis (range, 5–16). Diagnoses included polycystic ovarian syndrome, various forms of amenorrhea, premature ovarian insufficiency, vitamin and iron deficiency, hyperthyroidism and hypothyroidism, insulin sensitivity, and prolactinoma. CONCLUSION: An online clinical decision support tool can significantly reduce time to diagnosis of conditions affecting female fertility. It also reduces the number of patient–physician interactions and the time physicians spend on medical history and necessary patient education.

Determination of the Cancer Genome Atlas (TCGA) Endometrial Cancer Molecular Subtypes Using the Variant Interpretation and Clinical Decision Support Software MH Guide

The Cancer Genome Atlas (TCGA) network (United States National Cancer Institute) identified four molecular endometrial cancer (EC) subtypes using an extensive multi-method approach. The aim of this study was to determine the four TCGA EC molecular subtypes using a single-method whole-exome sequencing (WES)-based approach provided by MH Guide (Molecular Health, Heidelberg, Germany). WES and clinical data of n = 232 EC patients were obtained from TCGA. The four TCGA EC molecular subtypes designated as (i) Mutated Polymerase ε (POLE), (ii) Microsatellite Instability (MSI), (iii) Copy Number (CN) low and, (iv) CN-high were determined using the MH Guide software. The prognostic value of the subtypes determined by MH Guide were compared with the TCGA classification. Analysis of WES data using the MH Guide software led to the precise identification of the four EC molecular subtypes analogous to the TCGA classification. Both approaches displayed high concordance in terms of prognostic significance. The multi-method-based TCGA EC molecular subtypes can reliably be reproduced by the single-method-based MH Guide approach. The easy-to-implement single-method MH Guide approach represents a promising diagnostic tool.

FDA Regulation of Laboratory Clinical Decision Support Software: Is It a Medical Device?

On September 28, 2022, the Food & Drug Administration (FDA) released its guidance for clinical decision support (CDS) software (1), in which it outlines the criteria by which the FDA will determine whether a commercial CDS software will be regulated as a medical device (similar to a laboratory testing device) or be declared a “non-device” with lower regulatory burden. This distinction will impact the growing discipline of computerized CDS applied to laboratory testing, and it is critical for laboratory medicine professionals to be aware of and adapt to this rapidly changing field. CDS can be defined as “the use of information and communication technologies to bring relevant knowledge to bear on the healthcare and well-being of a patient” (2). CDS filters and organizes patient-specific data and presents it to caregivers to guide decisions regarding a patient’s health (3). CDS interventions have found increasing use in laboratory medicine, both to foster appropriate use of laboratory testing as well as to improve healthcare decisions based on test results (4, 5). Examples include avoiding order errors (6), preventing duplicate testing (7), and predicting outcomes or disease states using available test results (8–10).

The utilization of clinical decision support tools to identify neonatal hypothermia and its associated risk factors: A prospective observational study.

Hypothermia (axillary temperature less than 36.5°) is a major source of neonatal morbidity and mortality, with a disproportionate burden of disease in low- and middle-income countries. Despite the importance of thermoregulation on newborn outcomes, the global epidemiologic landscape of neonatal hypothermia is poorly characterized. Clinical decision support (CDS) software provides point-of-care recommendations to guide clinical management and may support data capture in settings with limited informatics infrastructure. Towards this end, we conducted a prospective observational study of the NoviGuide, a novel CDS platform for newborn care, at four health facilities in Uganda between September 2022 to May 2021. Data were extracted from clinical information (e.g. axillary temperature, birth weight, gestational age) entered into the NoviGuide by nurses and midwives on newborns within 24 hours of delivery. Descriptive statistics and multivariable logistic regression were used to evaluate neonatal temperature profiles and the association between hypothermia and clinical features. Among 1,027 completed assessments, 30.5% of entries had neonatal hypothermia with significant variation across study sites. On multivariable logistic regression analysis, we found that hypothermia was independently associated with pre-term birth (Adjusted Odd's Ratio [aOR] 2.62, 95% Confidence interval [CI] 1.38-4.98), sepsis/concern for sepsis (aOR 2.73, 95% CI 2.90-3.94), and hypoglycemia/concern for hypoglycemia (aOR 1.78, 95% CI 1.17-2.72). Altogether, neonatal hypothermia was commonly entered into the NoviGuide and associated clinical characteristics aligned with previous studies based on conventional data collection instruments. Our results should be contextualized within unique technical and operational features of CDS tools, including a bias towards acutely ill patients and limited quality control. Nonetheless, this study demonstrates that a CDS used voluntarily by clinicians has the potential to fill key data gaps and drive quality improvement towards reducing neonatal hypothermia in low resource settings.

Essential Regulatory Update for Implementation of Clinical Decision Support Software in the Psychiatric Field

Decision-making in the field of psychiatry, including diagnosis, is highly complex and is hindered by use of a dichotomous categorical classification system that is ill-suited to mental disorders which are multi-factorial behavioral conditions. In addition, traditional use of results from randomized controlled trials to guide medical device regulation in the field of psychiatry is also problematic, as marked heterogeneity exists within psychiatric patient groups, which precludes results from these trials being projected onto the general patient population. In the past 20 years, clinical decision support software (CDSS) has been found to improve decision-making abilities, but the traditional regulatory approach based on the categorical classification system and randomized controlled trials does not allow the necessary flexibility for CDSS-based decision-making in psychiatry. In this article, we will use Major Depressive Disorder as an example and will discuss regulatory considerations for CDSS, including artificial intelligence, in psychiatry. We will also provide an adjusted life-cycle framework for CDSS in psychiatry, given that the particular complexity of psychiatric disorders demands new and innovative decision support tools. We suggest that any new software would need to perform at least as well as the standard-of-care, which in psychiatry is an unfortunate trial-and-error process. This would be demonstrated during the pre-market validation stage using clinical data from back-end testing of the CDSS. We propose that pre-market evidence of CDSS efficacy should be based on parameters that are used to measure the software success rate, with evidence of safety including demonstration of the low risk of CDSS due to human involvement in the decision-making process. In the post-market stage, CDSS would be used by doctors to generate real-world data that would allow ongoing evaluation and improvement of the algorithms. Furthermore, CDSS would collect data beyond the initial intended-use patient population, allowing the CDSS to learn about related indications. These data would inform the pre-market phase, during which the CDSS could be updated with an expanded patient population. We anticipate that such changes would support effective use of CDSS in psychiatry and improved patient care, which is particularly important given the trial-and-process that comprises the current standard-of-care in the field.

Food and Drug Administration Published Final Guidance on Clinical Decision Support Software

Corresponding author: Jay Geller, JD, MBA, 121000 Wilshire Blvd, Suite 800, Los Angeles, CA 20025 ([email protected]). The author declares no conflicts of interest.

A clinical decision support software for the selection of pharmacological treatment in depression

A clinical decision support software for the selection of pharmacological treatment in depression