Clinical Decision Aid Research Articles

A tool for safer prescribing in vulnerable adults: the continuing development of the Medichec app and website.

Adverse effects are a common concern when prescribing and reviewing medication, particularly in vulnerable adults such as older people and those with intellectual disability. This paper describes the development of an app giving information on side-effects, called Medichec, and provides a description of the processes involved in its development and how drugs were rated for each side-effect. Medications with central anticholinergic action, dizziness, drowsiness, hyponatraemia, QTc prolongation, bleeding and constipation were identified using the British National Formulary (BNF) and frequency of occurrence of these effects was determined using the BNF, product information and electronic searches, including PubMed. Medications were rated using a traffic light system according to how commonly the adverse effect was known to occur or the severity of the effect. Medichec can facilitate access to side-effects information for multiple medications, aid clinical decision-making, optimise treatment and improve patient safety in vulnerable adults.

Applying clinical decision aids for the assessment and management of febrile infants presenting to emergency care in the UK and Ireland: Febrile Infant Diagnostic Assessment and Outcome (FIDO) Study protocol

IntroductionFebrile infants 90 days and younger are at risk of invasive bacterial infections (bacteraemia and meningitis) and urinary tract infections. Together this is previously termed serious bacterial infection with an...

QGenome: a smart app-based clinical decision aid for inherited cardiac conditions

Abstract Funding Acknowledgements Type of funding sources: Public Institution(s). Main funding source(s): Guy’s and St Thomas’ Charity South East Genomic Medicine Service Alliance Background QGenome is a smart application designed by a Regional Genetics service, in partnership with a Genomic Medicine Service Alliance (GMSA) and a technology company. The aims were to facilitate streamlined genomic testing, risk assessment, and referral guidance to relevant healthcare professionals in primary, secondary, and tertiary care settings; and to provide equity of "Genomics care" for patients across the UK, through standardized genomic testing and referral pathways. Due to high clinical demand, and following a design workshop with stakeholders, the initial version of QGenome was implemented for Cancer Genomics pathways. The scope has been extended to include workflows for Cardiology and Renal Genomics. The current Inherited Cardiac Conditions (ICC) workflow includes initial genomic testing/risk assessment pathways developed for familial hypercholesterolaemia, long QT syndrome, and vascular Ehlers Danlos Syndrome (vEDS). Methods For each clinical indication/condition, QGenome provides workflow assessments to determine eligibility for genomic testing and/or specialist tertiary referral. The workflow takes the user through a series of questions about their patient’s personal, past medical, and family history in a user-friendly format, to assess inherited susceptibility to disease. Eligibility criteria are derived from the UK National Genomic Test Directory, published literature and guidance from relevant advisory board working groups. The app can be used to find relevant forms, save reports, and activate referrals to specialist services or genetics clinics. QGenome is freely available on iOS, android and web-based platforms. Results The ICC workflows are demonstrated through the vEDS case studies (Figure). Conclusion QGenome offers a quick, efficient clinical decision tool for healthcare professionals, including cardiovascular nurses and allied professionals, in busy clinical settings, to differentiate between patients with likely sporadic cardiac disease and those with probable genetic/hereditary susceptibility for ICC.

Development and validation of a prognostic nomogram for the renal relapse of lupus nephritis

ObjectivesThis study aims to assess the risk of relapse after complete remission (CR) and partial remission (PR), and to develop a prognostic nomogram predicting the probability in lupus nephritis (LN) patients. MethodsData from patients with LN who had been in remission were collected as a training cohort. The prognostic factors were analyzed using the univariable and multivariable Cox model for the training group. A nomogram was then developed using significant predictors in multivariable analysis. Both discrimination and calibration were assessed by bootstrapping with 100 resamples. ResultsA total of 247 participants were enrolled, including 108 in the relapse group and 139 in the no relapse group. In multivariate Cox analysis, Systemic Lupus Erythematosus Disease Activity Index (SLEDAI), erythrocyte sedimentation rate (ESR), complement 1q (C1q), and antiphospholipid (aPL), anti-Sm antibody were found to be significant for predicting relapse rates. The prognostic nomogram including the aforementioned factors effectively predicted 1- and 3-year probability of flare-free. Moreover, a favorable consistency between the predicted and actual survival probabilities was demonstrated using calibration curves. ConclusionsHigh SLEDAI, ESR, and positive aPL, anti-Sm antibody are potential risk factors for LN flare, while high C1q can reduce its recurrence. The visualized model we established can help predict the relapse risk of LN and aid clinical decision-making for individual patients.

Novel Adaptive Histogram Binning-Based Lesion Segmentation for Discerning Severity in COVID-19 Chest CT Scan Images

Coronavirus sickness (COVID-19) recently adversely disrupted the medical care system and the entire economy. Doctors, researchers, and specialists are working on new-fangled methods to detect COVID-19 relatively efficiently, such as constructing computerized COVID-19 detection systems. Medical imaging, such as Computed Tomography (CT), has a lot of opportunity as a solution to RT-PCR approaches for quantitative assessment and disease monitoring. COVID-19 diagnosis based on CT images can provide speedy and accurate results. A quantitative criterion for diagnosis is provided by an automated segmentation method of infection areas in the lungs. As an outcome, automatic image segmentation is in high demand as a clinical decision aid tool. To detect COVID-19, Computed Tomography images might be employed instead of the time-consuming RT-PCR assay. In this research, a unique technique is provided for segmenting infection areas in the lungs using CT scan images from COVID-19 patients. “Ground Glass Opacity (GGO)” regions were detected using Novel Adaptive Histogram Binning Based Lesion Segmentation (NAHBLS) method. Many metrics were also employed to evaluate the proposed method, including “Sorensen–Dice similarity”, “Sensitivity”, “Specificity”, “Precision”, and “Accuracy” measures. Experiments have shown that the proposed method can effectively separate the lung infections with good accuracy. The results show that the proposed Novel Adaptive Histogram Binning Based Lesion Segmentation based on automatic approach is effective at segmenting the lesion region of the image and calculated the Infection Rate (IR) over the lung region in Computed Tomography scan.

Using chemotherapy response by KELIM score to predict response to first line maintenance PARP inhibitor therapy in non-BRCA mutant/homologous recombination deficiency (HRD) unknown high grade serous ovarian cancer (HGSOC).

e17547 Background: Maintenance PARP inhibitor therapy after response to first line chemotherapy is now standard of care in advanced HGSOC. Niraparib is available to all patients based on the PRIMA trial data; however, the progression free survival (PFS) benefit for patients without BRCA mutations or homologous recombination deficiency (HRD) is limited. Funded HRD testing is not accessible in many countries. Patient selection for PARPi therapy in non-BRCA mutant HGSOC is challenging. The calculated CA-125 ELIMination of Rate Constant K (KELIM) score is a mathematical model developed to evaluate CA-125 kinetics during chemotherapy. KELIM has been shown to correlate with chemosensitivity, with scores ≥1 associated with better clinical outcomes. This project aims to use real-world patient data to assess if surrogate markers, such as KELIM score and/or pathological chemotherapy response score, can predict response to 1st line maintenance PARPi in the absence of funded HRD testing. Methods: A retrospective review of non-BRCA mutant HGSOC cases on first line maintenance PARPi therapy at BC Cancer, Canada between April 2020 and June 2022. Only cases confirmed to be non-BRCA mutant (by both germline and tumour testing) were included in the study. Data collection was through electronic medical records and included patient demographics, chemotherapy intent (neoadjuvant vs. adjuvant), pathological (p) chemotherapy response score (CRS), progression-free survival (PFS) defined as start of niraparib to radiological evidence of disease progression. PFS analysis was performed in all patients provided 1 full cycle had been completed. The rate of not progressing at 12 months (PFS12) was calculated. KELIM score was calculated from at least 3 CA125 values taken within 100 days from the chemotherapy start date and using a validated calculation software. Results: 70 patients met the full eligibility criteria for analysis. All patients received niraparib as PARPi therapy. Mean age was 67 years and 40 patients (57%) were ≥ 65 years. Most patients were FIGO stage 3C at presentation (56%). Median number of niraparib cycles was 10 (range 1-28). 35/70 patients (50%) had disease progression at time of data analysis with a median follow-up of 13.2 months. 59 patients had evaluable KELIM scores. pCRS was not associated with any statistically significant differences in PFS. Patients with KELIM scores ≥1 had a trend to greater PFS from niraparib vs. those with KELIM <1 with median PFS of 15 months vs. 8.3 months respectively ( p=0.06). PFS12 rate was higher at 64% with KELIM scores ≥1 vs. 43% with KELIM <1; ( p=0.18). Conclusions: We show that KELIM score could be a useful tool to predict PARP inhibitor response and aid clinical decision-making by oncologists and patients in the real world setting where HRD testing is unfunded.

Predictive value of clinician impression for readmission and postdischarge mortality among neonates and young children in Dar es Salaam, Tanzania and Monrovia, Liberia

BackgroundThere are no validated clinical decision aids to identify neonates and young children at risk of hospital readmission or postdischarge mortality in sub-Saharan Africa, leaving the decision to discharge a...

A signature of cuproptosis-related lncRNAs predicts prognosis and provides basis for future anti-tumor drug development in breast cancer.

Breast cancer is the most prevalent malignancy worldwide and the leading culprit for women's death. Cuproptosis is a novel and promising modality of tumor cell death and the relationship with long non-coding RNAs (lncRNAs) remains shrouded in a veil. Studies in cuproptosis-related lncRNAs can aid in the clinical management of breast cancer and provide a basis for anti-tumor drug development. RNA-Seq data, somatic mutation data, and clinical information were downloaded from The Cancer Genome Atlas (TCGA). Patients were divided into high- and low-risk groups according to the risk score. Cox regression and least absolute shrinkage and selection operator (LASSO) regression analyses were used to select prognostic lncRNAs to construct a risk score system. Its' prognostic value was confirmed in the training and validation cohorts subsequently. Functional analysis regarding cuproptosis-related lncRNAs was performed. Eighteen cuproptosis-related lncRNAs were identified and 11 of them including AL023882.1, AC091588.1, AC138028.2, AC027514.1, AL592301.1, LRRC8C-DT, MFF-DT, NIFK-AS1, MECOM-AS1, OTUD6B-AS1 and RNF32-AS1 were selected for risk score system construction. The risk score was confirmed as an independent prognostic factor and patients in the high-risk group had a worse prognosis. A nomogram based on the independent prognostic factors was constructed for clinical decision aids. Further analyses revealed that patients in the high-risk group faced a heavier tumor mutational burden (TMB) and suppressed anti-tumor immunity. Besides, cuproptosis-related lncRNAs were associated with the expression of immune checkpoint inhibitors, N6-adenylate methylation (m6a), and drug sensitivity in breast cancer. A prognostic risk score system with satisfactory predictive accuracy was constructed. Besides, cuproptosis-related lncRNAs can influence the immune microenvironment, TMB, m6a, and drug sensitivity in breast cancer, which may provide a basis for future anti-tumor drug development.

Evaluation of the Canadian Clinical Practice Guidelines Risk Prediction Tool for Acute Aortic Syndrome: The RIPP Score.

Acute aortic syndrome (AAS) is a rare clinical syndrome with a high mortality rate. The Canadian clinical practice guideline for the diagnosis of AAS was developed in order to reduce the frequency of misdiagnoses. As part of the guideline, a clinical decision aid was developed to facilitate clinician decision-making (RIPP score). The aim of this study is to validate the diagnostic accuracy of this tool and assess its performance in comparison to other risk prediction tools that have been developed. This was a historical case-control study. Consecutive cases and controls were recruited from three academic emergency departments from 2002-2020. Cases were identified through an admission, discharge, or death certificated diagnosis of acute aortic syndrome. Controls were identified through presenting complaint of chest, abdominal, flank, back pain, and/or perfusion deficit. We compared the clinical decision tools' C statistic and used the DeLong method to test for the significance of these differences and report sensitivity and specificity with 95% confidence intervals. We collected data on 379 cases of acute aortic syndrome and 1340 potential eligible controls; 379 patients were randomly selected from the final population. The RIPP score had a sensitivity of 99.7% (98.54-99.99). This higher sensitivity resulted in a lower specificity (53%) compared to the other clinical decision aids (63-86%). The DeLong comparison of the C statistics found that the RIPP score had a higher C statistic than the ADDRS (-0.0423 (95% confidence interval -0.07-0.02); P < 0.0009) and the AORTAs score (-0.05 (-0.07 to -0.02); P = 0.0002), no difference compared to the Lovy decision tool (0.02 (95% CI -0.01-0.05 P < 0.25)) and decreased compared to the Von Kodolitsch decision tool (0.04 (95% CI 0.01-0.07 P < 0.008)). The Canadian clinical practice guideline's AAS clinical decision aid is a highly sensitive tool that uses readily available clinical information. It has the potential to improve diagnosis of AAS in the emergency department.

A treatment prediction strategy for overactive bladder using a machine learning algorithm that utilized data from the FAITH study.

To use machine learning algorithms to develop a model to accurately predict treatment responses to mirabegron or antimuscarinic agents in patients with overactive bladder (OAB), using real-world data from the FAITH registry (NCT03572231). The FAITH registry data included patients who had been diagnosed with OAB symptoms for at least 3 months and were due to initiate monotherapy with mirabegron or any antimuscarinic. For the development of the machine learning model, data from patients were included if they had completed the 183-day study period, had data for all timepoints and had completed the overactive bladder symptom scores (OABSS) at baseline and end of study. The primary outcome of the study was a composite outcome combining efficacy, persistence, and safety outcomes. Treatment was deemed "more effective" if the composite outcome criteria for "successful," "no treatment change," and "safe" were met, otherwise treatment was deemed "less effective." To explore the composite algorithm, a total of 14 clinical risk factors were included in the initial data set and a 10-fold cross-validation procedure was performed. A range of machine learning models were evaluated to determine the most effective algorithm. In total, data from 396 patients were included (266 [67.2%] treated with mirabegron and 130 [32.8%] treated with an antimuscarinic). Of these, 138 (34.8%) were in the "more effective" group and 258 (65.2%) were in the "less effective" group. The groups were comparable in terms of their characteristic distributions across patient age, sex, body mass index, and Charlson Comorbidity Index. Of the six models initially selected and tested, the decision tree (C5.0) model was chosen for further optimization, and the receiver operating characteristic of the final optimized model had an area under the curve result of 0.70 (95% confidence interval: 0.54-0.85) when 15 was used for the min n parameter. This study successfully created a simple, rapid, and easy-to-use interface that could be further refined to produce a valuable educational or clinical decision-making aid.

Influence of adverse effects of neoadjuvant chemoradiotherapy on the prognosis of patients with early-stage esophageal cancer (cT1b-cT2N0M0) based on the SEER database.

To analyze the prognostic impact of neoadjuvant chemoradiotherapy (NCRT) on early-stage (cT1b-cT2N0M0) esophageal cancer (ESCA) and construct a prognostic nomogram for these patients. We extracted the clinical data about patients diagnosed with early-stage esophageal cancer from the 2004-2015 period of the Surveillance, Epidemiology, and End Results (SEER) database. We applied the independent risk factors affecting the prognosis of patients with early-stage esophageal cancer obtained after screening by univariate and multifactorial COX regression analyses to establish the nomogram and performed model calibration using bootstrapping resamples. The optimal cut-off point for continuous variables is determined by applying X-tile software. After balancing the confounding factors by propensity score matching (PSM) and inverse probability of treatment weighting (IPTW) method, Kaplan-Meier(K-M) curve, and log-rank test were applied to evaluate the prognostic impact of NCRT on early-stage ESCA patients. Among patients who met the inclusion criteria, patients in the NCRT plus esophagectomy (ES) group had a poorer prognosis for overall survival (OS) and esophageal cancer-specific survival (ECSS) than patients in the ES alone group (p < 0.05), especially in patients who survived longer than 1 year. After PSM, patients in the NCRT + ES group had poorer ECSS than patients in the ES alone group, especially after 6 months, while OS was not significantly different between the two groups. IPTW analysis showed that, prior to 6 months patients in the NCRT + ES group had a better prognosis than patients in the ES group, regardless of OS or ECSS, whereas after 6 months, patients in the NCRT + ES group had a poorer prognosis. Based on multivariate COX analysis, we established a prognostic nomogram which showed areas under the ROC curve (AUC) for 3-, 5-, and 10-year OS 0.707, 0.712, and 0.706, respectively, with the calibration curves showing that the nomogram was well calibrated. Patients with early-stage ESCA (cT1b-cT2) did not benefit from NCRT, and we established a prognostic nomogram to provide clinical decision aid for the treatment of patients with early-stage ESCA.

Influence of Patient and Clinician Gender on Emergency Department HEART Scores: A Secondary Analysis of a Prospective Observational Trial

Clinical decision aids can decrease health care disparities. However, many clinical decision aids contain subjective variables that may introduce clinician bias. The HEART score is a clinical decision aid that estimates emergency department (ED) patients' cardiac risk. We sought to explore patient and clinician gender's influence on HEART scores. In this secondary analysis of a prospective observational trial, we examined a convenience sample of adult ED patients at one institution presenting with acute coronary syndrome symptoms. We compared ED clinician-generated HEART scores with researcher-generated HEART scores blinded to patient gender. The primary outcome was agreement between clinician and researcher HEART scores by patient gender overall and stratified by clinician gender. Analyses used difference-in-difference (DiD) for continuous score and prevalence-adjusted, bias-adjusted Kappa (PABAK) for binary (low versus moderate/high risk) score comparison. All 336 clinician-patient pairs from the original study were included. In total, 47% (158/336) of patients were women, and 52% (174/336) were treated by a woman clinician. The DiD between clinician and researcher HEART scores among men versus women patients was 0.24 (95% CI-0.01 to 0.48). Compared with researchers, men clinicians assigned a higher score to men versus women patients (DiD 0.51 [95% CI 0.16 to 0.87]), whereas women clinicians did not (DiD 0.00 [95% CI-0.33 to 0.33]). Agreement was the highest among women clinicians (PABAK 0.72; 95% CI 0.61 to 0.81) and lowest among men clinicians assessing men patients (PABAK 0.47; 95% CI 0.29 to 0.66). Patient and clinician gender may influence HEART scores. Researchers should strive to understand these influences in developing and implementing this and other clinical decision aids.

A step in the right direction: the potential role of smartwatches in supporting chronic disease prevention in health care.

A step in the right direction: the potential role of smartwatches in supporting chronic disease prevention in health care.

MyopiaDETR: End-to-end pathological myopia detection based on transformer using 2D fundus images.

Automated diagnosis of various retinal diseases based on fundus images can serve as an important clinical decision aid for curing vision loss. However, developing such an automated diagnostic solution is challenged by the characteristics of lesion area in 2D fundus images, such as morphology irregularity, imaging angle, and insufficient data. To overcome those challenges, we propose a novel deep learning model named MyopiaDETR to detect the lesion area of normal myopia (NM), high myopia (HM) and pathological myopia (PM) using 2D fundus images provided by the iChallenge-PM dataset. To solve the challenge of morphology irregularity, we present a novel attentional FPN architecture and generate multi-scale feature maps to a traditional Detection Transformer (DETR) for detecting irregular lesion more accurate. Then, we choose the DETR structure to view the lesion from the perspective of set prediction and capture better global information. Several data augmentation methods are used on the iChallenge-PM dataset to solve the challenge of insufficient data. The experimental results demonstrate that our model achieves excellent localization and classification performance on the iChallenge-PM dataset, reaching AP50 of 86.32%. Our model is effective to detect lesion areas in 2D fundus images. The model not only achieves a significant improvement in capturing small objects, but also a significant improvement in convergence speed during training.

Implementierung einer elektronischen Entscheidungshilfe zur Unterstützung der Behandlungsqualität der COPD in der Hausarztmedizin

BackgroundElectronic clinical decision aids (eCDAs) have the potential to improve the quality of chronic disease management (CDM) and, therefore, patient relevant outcomes. However, eCDAs are only sparsely implemented in primary care for chronic obstructive pulmonary disease (COPD). The aim of this pilot study was to develop and implement of an eCDA for COPD primary care in two Swiss primary care practices. MethodTwo primary care practices, each with five primary care physicians and their assistants, participated in the study. The eCDA was developed in collaboration with one of the two GP practices (Practice 1) following a development cycle encompassing alpha- and beta-testing stages. Long-term testing for one year was conducted in both practices. The implementation of the eCDA was evaluated according to the RE-AIM framework counting occurrences as follows: Reach: the number of patients included in the CDM using the eCDA. Effectiveness: the number of treatment processes initiated per patient. Adoption: practice utilization of the diverse functions featured in the eCDA. Implementation and Maintenance: health care professionals’ attitudes towards the impact of the eCDA on the quality of care and their willingness to continue using the eCDA after long-term testing. Data were collected by the eCDA itself, which was programmed to track user data, and from practice staff using questionnaires. ResultsReach: After the long-term test, the number of patients recorded in the eCDA was 28 in practice 1, and 12 in practice 2. Effectiveness: The number of evidence-based treatment processes per patient was 14 (IQR 6 to 22) in Practice 1 and 6 (IQR 5 to 8) in Practice 2. Adoption: The utilization profiles of the eCDA differed greatly between practices. Implementation and Maintenance: After the long-term test, respondents were more critical of the quality of the CDM for patients with COPD, and attitudes consistent with interprofessional care were more prevalent compared to baseline. Respondents were optimistic regarding both the potential of the eCDA to improve the quality of CDM and their motivation to continue using the eCDA after long-term testing. ConclusionThis pilot study is a roadmap for future projects aiming to develop and implement eCDAs for the CDM of COPD in primary care. Future larger implementation studies in this domain should place greater emphasis on the measurement of structural practice characteristics as potential determinants of patient-relevant outcomes. The modifiable determinants should then be tested for their effects on patient-relevant outcomes in a randomized controlled design.

Patient similarity and other artificial intelligence machine learning algorithms in clinical decision aid for shared decision-making in the Prevention of Cardiovascular Toxicity (PACT): a feasibility trial design

BackgroundThe many improvements in cancer therapies have led to an increased number of survivors, which comes with a greater risk of consequent/subsequent cardiovascular disease. Identifying effective management strategies that can mitigate this risk of cardiovascular complications is vital. Therefore, developing computer-driven and personalized clinical decision aid interventions that can provide early detection of patients at risk, stratify that risk, and recommend specific cardio-oncology management guidelines and expert consensus recommendations is critically important.ObjectivesTo assess the feasibility, acceptability, and utility of the use of an artificial intelligence (AI)-powered clinical decision aid tool in shared decision making between the cancer survivor patient and the cardiologist regarding prevention of cardiovascular disease.DesignThis is a single-center, double-arm, open-label, randomized interventional feasibility study. Our cardio-oncology cohort of > 4000 individuals from our Clinical Research Data Warehouse will be queried to identify at least 200 adult cancer survivors who meet the eligibility criteria. Study participants will be randomized into either the Clinical Decision Aid Group (where patients will use the clinical decision aid in addition to current practice) or the Control Group (current practice). The primary endpoint of this study is to assess for each patient encounter whether cardiovascular medications and imaging pursued were consistent with current medical society recommendations. Additionally, the perceptions of using the clinical decision tool will be evaluated based on patient and physician feedback through surveys and focus groups.SummaryThis trial will determine whether a clinical decision aid tool improves cancer survivors’ medication use and imaging surveillance recommendations aligned with current medical guidelines.Trial registrationClinicalTrials.Gov Identifier: NCT05377320

Monkeypox Classification based on Skin Images using CNN: EfficientNet-B0

Monkeypox is a zoonotic infectious disease caused by a virus of the orthopoxvirus genus. It can infect humans, vertebrates, and arthropods. Transmission to humans occurs through direct contact with infected animal body fluids or consumption of undercooked meat. Monkeypox cases have been reported globally, with thousands of confirmed cases and several deaths. Early symptoms include fever, rash, swollen lymph nodes, back pain, and headache. Diagnosis can be made through physical examination and laboratory tests. Imagebased artificial intelligence technology, specifically the EfficientNet-B0 architecture, has been proposed as a solution for the classification of monkeypox based on skin lesion images. The research aims to compare the performance of EfficientNetB0 with other CNN architectures and contribute to the development of medical image classification technology. Among the models evaluated, the EfficientNet-B0 model emerged as the standout performer, achieving an accuracy of 85.12%, surpassing the accuracy of other models such as MobileNet (63.63%) and InceptionV3 (71.4%). EfficientNet-B0 also demonstrated strong sensitivity (78.46%) and impressive specificity (91.78%), outperforming other models in these metrics. Additionally, despite not surpassing the accuracy of ResNet-50 (87.59%), EfficientNet-B0 achieved its accuracy with approximately four times fewer parameters, highlighting its efficiency in parameter usage and computational resources. These results can help improve models and aid in clinical decision-making.

Time for More Support? Stakeholder Engagement for a Long-Term Care Planning Dementia Decision Aid Prototype.

Clinical decision aids around long-term care can help support persons living with dementia (PLWD), family care partners, and healthcare providers navigate current and future care decisions. This study describes the iterative development of a long-term care planning dementia decision aid and explores care partner and geriatric providers' insights regarding its acceptability and usability. Using a convergent parallel mixed methods design, we gathered surveys and completed interviews with 11 care partners and 11 providers. The quantitative and qualitative data were then converged, resulting in four findings: (1) helpfulness of the decision aid in supporting future care planning; (2) versatility of the decision aid in practice; (3) preferences for structure and content of the decision aid; and (4) perceived shortcomings of the decision aid in decision making. Future work should continue to refine the decision aid, pilot implementation, and evaluate potential effects on decision making as part of dementia care.

Predicting 30-day all-cause hospital readmission using multimodal spatiotemporal graph neural networks.

Reduction in 30-day readmission rate is an important quality factor for hospitals as it can reduce the overall cost of care and improve patient post-discharge outcomes. While deep-learning-based studies have shown promising empirical results, several limitations exist in prior models for hospital readmission prediction, such as: (a) only patients with certain conditions are considered, (b) do not leverage data temporality, (c) individual admissions are assumed independent of each other, which ignores patient similarity, (d) limited to single modality or single center data. In this study, we propose a multimodal, spatiotemporal graph neural network (MM-STGNN) for prediction of 30-day all-cause hospital readmission, which fuses in-patient multimodal, longitudinal data and models patient similarity using a graph. Using longitudinal chest radiographs and electronic health records from two independent centers, we show that MM-STGNN achieved an area under the receiver operating characteristic curve (AUROC) of 0.79 on both datasets. Furthermore, MM-STGNN significantly outperformed the current clinical reference standard, LACE+ (AUROC=0.61), on the internal dataset. For subset populations of patients with heart disease, our model significantly outperformed baselines, such as gradient-boosting and Long Short-Term Memory models (e.g., AUROC improved by 3.7 points in patients with heart disease). Qualitative interpretability analysis indicated that while patients' primary diagnoses were not explicitly used to train the model, features crucial for model prediction may reflect patients' diagnoses. Our model could be utilized as an additional clinical decision aid during discharge disposition and triaging high-risk patients for closer post-discharge follow-up for potential preventive measures.

Brain structure and stroke risk score in subjects without a history of atrial fibrillation

CHA2DS2-VASc score system aids in clinical decision-making in subjects with atrial fibrillation (AF). Little is known on the association between CHA2DS2-VASc scores and brain structure in patients without cardiac arrhythmia. Detailed brain architecture analysis was performed. Assessment of bivariate correlation between the volume of segmented brain structures and Z-scores of CHA2DS2-VASc showed that higher risk scores correlated negatively and significantly with various brain framework. Our study confirms that a cluster of risk factors incorporated in a well-established risk score correlated with brain tissue volume independently of the presence of an arrhythmia.