Exploring telediagnostic procedures in child neuropsychiatry: addressing ADHD diagnosis and autism symptoms through supervised machine learning

Recently, there has been an increase in telemedicine applied to child neuropsychiatry, such as the use of online platforms to collect remotely case histories and demographic and behavioral information. In the present proof-of-concept study, we aimed to understand to what extent information parents and teachers provide through online questionnaires overlaps with clinicians' diagnostic conclusions on attention-deficit/hyperactivity disorder (ADHD). Moreover, we intended to explore a possible role that autism spectrum disorders (ASD) symptoms played in this process. We examined parent- and teacher-rated questionnaires collected remotely and an on-site evaluation of intelligence quotients from 342 subjects (18% females), aged 3-16 years, and referred for suspected ADHD. An easily interpretable machine learning model-decision tree (DT)-was built to simulate the clinical process of classifying ADHD/non-ADHD based on collected data. Then, we tested the DT model's predictive accuracy through a cross-validation approach. The DT classifier's performance was compared with those that other machine learning models achieved, such as random forest and support vector machines. Differences in ASD symptoms in the DT-identified classes were tested to address their role in performing a diagnostic error using the DT model. The DT identified the decision rules clinicians adopt to classify an ADHD diagnosis with an 82% accuracy rate. Regarding the cross-validation experiment, our DT model reached a predictive accuracy of 74% that was similar to those of other classification algorithms. The caregiver-reported ADHD core symptom severity proved the most discriminative information for clinicians during the diagnostic decision process. However, ASD symptoms were a confounding factor when ADHD severity had to be established. Telehealth procedures proved effective in obtaining an automated output regarding a diagnostic risk, reducing the time delay between symptom detection and diagnosis. However, this should not be considered an alternative to on-site procedures but rather as automated support for clinical practice, enabling clinicians to allocate further resources to the most complex cases.

Use of machine learning on clinical questionnaires data to support the diagnostic classification of Attention DeficitHyperactivity Disorder: a personalized medicine approach

Introduction

Attention Deficit / Hyperactivity Disorder (ADHD) is a highly prevalent neurodevelopmental condition characterized by inattention, motor hyperactivity and impulsivity. ADHD cognitive and behavioral presentation is characterized by a high heterogeneity (APA, 2013). Indeed, a complex diagnostic process, that considers several validated tools, is, to date, necessary.

Objectives

The main aim is to develop supervised machine learning (ML) algorithms that could be used to support the diagnostic process for ADHD, by identifying the most relevant features in discriminating between the presence or absence of the ADHD diagnosis in children.

Methods

We analyzed data from 342 children (Mean age: 8y 8m ± 1y; 61 F) referred for possible ADHD symptomatology. Assessments were performed by an expert clinician and through questionnaires: Social Responsiveness Scale (SRS), Child Behavior Checklist (CBCL), Conners Rating Scale for Parents (CPRS) and for Teachers (CTRS). Data were analyzed using a decision tree classifier and random forest algorithms.

Results

The decision tree model performed an accuracy of 0.71. The random forest model that was identified as the best tested, performed an accuracy of 0.77 (Figure 1) and it identified as most informative parent- and teacher-rated DSM-oriented ADHD symptoms (Figure 2).

Figure 1: Random forest confusion matrix and statistics.

Figure 2: Ranking of variables importance.

Conclusions

A random forest classifier could represent an effective algorithm to support the identification of ADHD children and to simplify the diagnostic process as an initial step. The use of supervised machine learning algorithms could be useful in helping the diagnostic process, highlighting the importance of a personalized medicine approach.

Disclosure

No significant relationships.

Characteristics and Outcomes of Recipients of Heart Transplant with Coronarvirus Disease 2019 Who Received Casirivimab Plus Imdevimab Infusion

Purpose

Heart transplant (HT) recipient are at increased risk of adverse outcomes following COVID-19 infection and may benefit from monoclonal antibody infusion to mitigate progression to clinically severe disease. The aim of this study is to describe the outcomes of HT patients who experienced mild to moderate coronavirus disease 2019 (COVID-19), with subsequent administration of casirivimab plus imdevimab administration.

Methods

A retrospective review of all HT recipients who were infected with COVID-19, and subsequently infused with monoclonal antibodies in a large academic medical center between January 1, 2021 to September 1, 2021.

Results

14 HT patients were included in the analysis. The median age was 57.5 (interquartile range [IQR], 41.5-64) years, 10 (71%) were men, and median time from HT was 3.48 (IQR, 1.00-11.82) years. Comorbid conditions included hypertension in 6 patients (43%), diabetes in 4 (29%), and chronic kidney disease in 6 (43%). Eight patients (57%) were previously vaccinated, predominantly with the Pfizer-BioNTech vaccine. Three participants (21%) were admitted after clinical progression of COVID-19. Among patients managed at the study institution, mycophenolate mofetil was discontinued in two patients (14%) and calcineurin inhibitor was maintained at previous levels in all fourteen patients (100%). Of the admitted patients, 1 was treated with high dose corticosteroids alone and 2 were treated with corticosteroids plus remdesivir. No patient required intubation. All 3 patients were discharged home and no patients in this cohort died.

Conclusion

In this single-center case series, HT patients with mild-moderate COVID-19 who were treated with monoclonal antibody infusion had a hospitalization rate of 21% and 100% survival. Further studies are required to optimize management of COVID-19 infection in the HT population.

Understanding feeding problems in autistic children: Exploring the interplay between internalizing symptoms and sensory features

Feeding/behavior problems are highly prevalent in autistic children. This study aimed to explore the interplay between autistic features, sensory processing patterns, emotional/behavioral difficulties, and feeding problems in children with a clinical diagnosis of autism spectrum disorder. Ninety-nine children aged 1.5–5 years were assessed using the Children’s Eating Behavior Inventory (CEBI), the Child Behavior Checklist, the Revised Sensory Profile—Second Edition (SP-2), and the Food Frequency Questionnaire. The sample was divided into two subgroups based on the presence of feeding behavior problems as rated by parents on the CEBI. Children with feeding problems showed more severe internalizing symptoms and were more reactive to sensory stimuli than children without feeding problems. We also found a significant relationship between higher levels of autistic features and greater feeding behavior problems. A mediation analysis model indicated that this relationship was mediated by sensory processing, as reflected by SP-2 quadrants scores. Although preliminary, these findings suggest that children with higher levels of autistic features could be at greater risk of developing feeding behavior problems only if they also have a higher sensory sensitivity to environmental stimuli.

Lay abstract

This study adds to a growing body of research documenting feeding problems in autistic children. Our results indicate that children aged 1.5–5 years with feeding problems may present with elevated internalizing difficulties and alterations in sensory processing when compared to same-age children without feeding problems. Our study also proposes that sensory processing may be an important, mediating factor in the relationship between autism features and feeding problems. The present work suggests, therefore, that implementers should thoroughly consider the sensory profile of autistic children prior to intervening on feeding behaviors. In particular, based on these preliminary findings, feeding interventions could benefit from environments designed to support children who can find them noisy and overwhelming.

MedicalBIT: A web platform for Standardized Data Acquisition, Processing and Export in Child Psychopathology Clinical Routine. From design to implementation. (Preprint)

Background

The rapid extent of digital innovation for the collection of data has transformed the way in which health professionals collect, share, and analyze health information for better clinical decision-making and health care. In the last decade, there has been an increased interest in telemedicine by mental health agencies; the gap between the need for care and both diagnosis and treatment is wide, and digital technology could play an important role in filling this gap. However, there are limited data on the effectiveness of the clinical process and cost-effectiveness of most telemedicine applications.

Objective

This study examined the implementation of the first Italian online, web-based, comprehensive screening tool and described the screening and diagnostic process through the interactive web platform in a child psychopathology clinic. This is a feasibility study that aims to present the design and implementation of the best practices to improve patient experiences and clinical outcomes. Moreover, the paper evaluates the platform with qualitative and quantitative measures.

Methods

We planned, designed, and implemented a web-based system to collect, store, and manage clinical data. The platform was developed by a multidisciplinary team composed of researchers, clinicians, and informatics professionals through different steps. First, we defined the clinical information to be collected. A number of measures were chosen, tapping several clinical risk areas such as neurodevelopmental disorders and emotional and behavioral problems. The web application architecture and process were then designed. The three phases of process design are described in detail: design of the input interface, processing design, and design of the output interface. Finally, the system has been implemented and evaluated. Based on indicators recommended by the National Quality Forum and the Italian National Guidelines, we evaluated the quality of the system and used quantitative measures that were replicable and comparable over time.

Results

We present the implemented architecture and features of Medea Information and Clinical Assessment On-Line (MedicalBIT), and we provide performance measures for the data collected between October 2018 and June 2021. The measured concepts pertain to four domains: access to care, financial impact/cost, experience, and effectiveness.

Conclusions

In this study, we present the successful implementation of an innovative digital tool. The findings of this study show that the implemented web-based platform appears to be an efficient, cost-effective, and feasible way to improve digital care in the field of child psychiatry.

Major adverse cardiovascular events associated with VEGF-targeted anticancer tyrosine kinase inhibitors: a real-life study and proposed algorithm for proactive management

BACKGROUND

Vascular endothelial growth factor receptor (VEGFR)-targeted tyrosine kinase inhibitors (TKIs) are widely used in cancer treatment and burdened by cardiovascular toxicity. The majority of data come from clinical trials, thus in selected populations. The aim of our study is to evaluate the cardiotoxicity profile of VEGFR-targeted TKIs and the impact of cardiovascular risk factors in a real-life population.

PATIENTS AND METHODS

In this cohort, population-based study, patients treated with VEGFR-targeted TKIs, bevacizumab and trastuzumab between 2009 and 2014 were analyzed. A multi-source strategy for data retrieval through hospital, pharmaceutical and administrative databases of the Lombardy region, Italy, has been adopted. The primary endpoint was to determine the incidence and type of major adverse cardiovascular events (MACEs) along with their temporal trend. The secondary endpoint was to define the impact of cardiovascular risk factors in the occurrence of MACEs.

RESULTS

A total of 829 patients were treated with VEGFR-targeted TKIs. Eighty-one MACEs occurred in the first year of follow-up [crude cumulative incidence (CCI): 9.79%] mainly consisting of arterial thrombotic events (ATEs, 31 events, CCI: 3.99%), followed by rhythm disorders (22 events, CCI: 2.66%), pulmonary embolisms and heart failures (13 events each, CCI: 1.57%). While the incidence of most MACEs showed a plateau after 6 months, ATEs kept increasing along the year of follow-up. Hypertension and dyslipidemia were associated with an increase in risk of ATEs [relative risk difference (RRD) +209.8% and +156.2%, respectively], while the presence of previous MACEs correlated with a higher risk of all MACEs in multivariate analysis (RRD 151.1%, 95% confidence interval 53.6% to 310.3%, P < 0.001).

CONCLUSIONS

MACEs occur in a clinically significant proportion of patients treated with VEGFR-targeted TKIs, with ATEs being predominant, mainly associated with hypertension and dyslipidemia. A clinical algorithm for effective proactive management of these patients is warranted.

Remote Monitoring of Heart Transplant Recipients during the COVID-19 Pandemic

Purpose

The COVID-19 pandemic created significant challenges in monitoring heart transplant (HT) recipients for rejection due to efforts to minimize contact with the hospital setting. The aim of this study was to evaluate the safety and efficacy of transitioning HT patients to home phlebotomy and a monitoring protocol based on gene expression profiling (GEP) and donor derived cell free DNA (ddcfDNA).

Methods

A single-center cohort study that prospectively enrolled consecutive HT patients who were transitioned to a remote monitoring protocol employing home phlebotomy and non-invasive surveillance for rejection. Patients were enrolled starting at 2 months post-HT. Positive GEP values were defined as ≥32 (up to 6 months post-HT) and ≥34 (> 6 months post-HT). A positive ddcfDNA score was defined as >0.12%. A positive biopsy was defined as grade ≥1B/1R

Results

246 HT patients were enrolled and followed for a minimum of 3 months. Mean age was 56±14, 71.5% were male, and median time from transplant was 2.7 years. The average distance of patients from the hospital was 25.6 miles. 359 blood tests were drawn for detection of GEP and ddcfDNA and 102 biopsies performed (Figure). Among 32 patients who had negative results on both tests and had a biopsy, 0 had a positive biopsy. Of 25 patients who had positive results on both tests and had a biopsy, 3 (12%) had a positive biopsy. The biopsy positivity rate in patients who were GEP+/ddcfDNA- was 6% and in patients who were GEP-/ddcfDNA+ was 8%. None of the positive biopsies were associated with hemodynamic compromise. 15 (6%) of patients were admitted due to allograft rejection during the study period. There were no deaths.

Conclusion

Using a remote monitoring protocol with home phlebotomy and noninvasive rejection surveillance was feasible and safe in HT recipients. In this cohort, the combination of negative GEP and ddcfDNA scores was accurate at predicting a lack of allograft rejection.

Quality Assessment of HTA reports: a Critical Appraisal Toolkit within an institutional HTA program

Issue

With population aging, increased demand for healthcare and limited availability of economic and human resources, well-functioning and sustainable health systems have to rely on rigorous and evidence-based assessments of complex technologies' clinical effectiveness and safety. The Lombardy Region in Northern Italy has a well-established HTA program that offers technical support to its hospitals' network to produce and to review Health Technology Assessment (HTA) reports.

Description of the Problem

To better coordinate the HTA network and to reduce self-referencing of University and Community hospitals, Lombardy Region carried out an intense field training and distance learning from year 2017, with a project aimed at involving health care professionals in an accurate evaluation of technologies in 18 HTA reports. The regional HTA Supporting Centre developed a Toolkit for the critical appraisal of reports and supporting literature to improve the quality of hospital-based reports.

Results

In compliance with the regional Law DGR XI/1046 17.12.2018 and the framework proposed, during year 2019 hospitals used the Toolkit to help writing complete and good quality HTA reports on 37 different technologies. With a Public Health resident internship, the Toolkit was revised, extended and then used in 2020 by the hospital's HTA network during the double blinded peer review, mandated by the regional Law DGR XI/2672 16.12.2019, of the year 2019 HTA reports. Detailed results will be presented at the conference.

Lessons

We observed an increase in number of HTA reports from 2018 to 2019, a greater language's and format's homogeneity and an improvement in the quality of some reports, which will be submitted to the Regional HTA Committee for appraisal and reimbursement.

Key messages

The development and use of a Critical Appraisal Toolkit in an Regional HTA program can help hospitals write complete and good quality HTA reports. A centrally supported quality improvement of distributed assessment activities in a network of hospitals can enhance the production of HTA reports, relevant to the needs of a local healthcare system.

Dual antiplatelet therapy prolongation in high-risk patients with prior myocardial infarction: insights from the post-PCI registry

BACKGROUND

Patients surviving a myocardial infarction (MI) are at a heightened risk for recurrent ischemic events that can be reduced with the long-term addition of a second antithrombotic drug to aspirin. However, data about real prescription of this therapy are lacking and sometimes controversial.

METHODS

We aimed to describe the incidence and the determinants of a dual antiplatelet therapy (DAPT) prolongation beyond 12 months in a cohort of consecutive patients undergoing percutaneous coronary intervention (PCI) with prior MI undergoing PCI and features of high ischemic risk intended as age more than 65 years, second MI, type 2 diabetes mellitus, multivessel coronary artery disease (MVCAD) and chronic kidney disease (CKD). We analysed patients enrolled in the prospective 'Post-PCI' registry that included patients treated with PCI for stable coronary artery disease (CAD) or acute coronary syndromes. At 12 months' follow-up, we collected data about DAPT prolongation in patients with prior MI and at least one of the previous features of high risk who did not experience ischemic and bleeding events during the follow-up.

RESULTS

Among 1113 patients included in the registry, 778 (72%) presented the inclusion criteria for the present study: 434 (66%) were more than 65 years old, 245 (37%) had a second MI, 189 (29%) diabetes mellitus, 480 (73%) MVCAD and 216 (33%) CKD. Despite a DAPT being prescribed for 1 year in 86% of the patients, it was prolonged for over 12 months in 105 (16%) of them. At multivariable analysis, only second MI and MVCAD were independent predictors of DAPT prolongation in a model including age more than 65 years, diabetes mellitus, CKD and PCI on left main/left anterior descending coronary artery. We found no significant difference in DAPT prolongation according to a DAPT-score value at least 2 or based on the physician who actually performed the follow-up (clinical cardiologist, interventional cardiologist or other).

CONCLUSION

In patients with prior MI and features of high ischemic risk undergoing PCI, the rate of DAPT prolongation beyond 12 months was low; recurrent MI and MVCAD appeared as its main determinants.