DrugDoctor: enhancing drug recommendation in cold-start scenario via visit-level representation learning and training.

Medication recommendation is a crucial application of artificial intelligence in healthcare. Current methodologies mostly depend on patient-level longitudinal representation, which utilizes the entirety of historical electronic health records for making predictions. However, they tend to overlook a few key elements: (1) The need to analyze the impact of past medications on previous conditions. (2) Similarity in patient visits is more common than similarity in the complete medical histories of patients. (3) It is difficult to accurately represent patient-level longitudinal data due to the varying numbers of visits. To our knowledge, current models face difficulties in dealing with initial patient visits (i.e. in cold-start scenarios) which are common in clinical practice. This paper introduces DrugDoctor, an innovative drug recommendation model crafted to emulate the decision-making mechanics of human doctors. Unlike previous methods, DrugDoctor explores the visit-level relationship between prescriptions and diseases while considering the impact of past prescriptions on the patient's condition to provide more accurate recommendations. We design a plug-and-play block to effectively capture drug substructure-aware disease information and effectiveness-aware medication information, employing cross-attention and multi-head self-attention mechanisms. Furthermore, DrugDoctor adopts a fundamentally new visit-level training strategy, aligning more closely with the practices of doctors. Extensive experiments conducted on the MIMIC-III and MIMIC-IV datasets demonstrate that DrugDoctor outperforms 10 other state-of-the-art methods in terms of Jaccard, F1-score, and PRAUC. Moreover, DrugDoctor exhibits strong robustness in handling patients with varying numbers of visits and effectively tackles "cold-start" issues in medication combination recommendations.

Auditing YouTube's recommendation system for ideologically congenial, extreme, and problematic recommendations.

Algorithms of social media platforms are often criticized for recommending ideologically congenial and radical content to their users. Despite these concerns, evidence on such filter bubbles and rabbit holes of radicalization is inconclusive. We conduct an audit of the platform using 100,000 sock puppets that allow us to systematically and at scale isolate the influence of the algorithm in recommendations. We test 1) whether recommended videos are congenial with regard to users' ideology, especially deeper in the watch trail and whether 2) recommendations deeper in the trail become progressively more extreme and come from problematic channels. We find that YouTube's algorithm recommends congenial content to its partisan users, although some moderate and cross-cutting exposure is possible and that congenial recommendations increase deeper in the trail for right-leaning users. We do not find meaningful increases in ideological extremity of recommendations deeper in the trail, yet we show that a growing proportion of recommendations comes from channels categorized as problematic (e.g., "IDW," "Alt-right," "Conspiracy," and "QAnon"), with this increase being most pronounced among the very-right users. Although the proportion of these problematic recommendations is low (max of 2.5%), they are still encountered by over 36.1% of users and up to 40% in the case of very-right users.

Precision oncology: a review to assess interpretability in several explainable methods.

Great efforts have been made to develop precision medicine-based treatments using machine learning. In this field, where the goal is to provide the optimal treatment for each patient based on his/her medical history and genomic characteristics, it is not sufficient to make excellent predictions. The challenge is to understand and trust the model's decisions while also being able to easily implement it. However, one of the issues with machine learning algorithms-particularly deep learning-is their lack of interpretability. This review compares six different machine learning methods to provide guidance for defining interpretability by focusing on accuracy, multi-omics capability, explainability and implementability. Our selection of algorithms includes tree-, regression- and kernel-based methods, which we selected for their ease of interpretation for the clinician. We also included two novel explainable methods in the comparison. No significant differences in accuracy were observed when comparing the methods, but an improvement was observed when using gene expression instead of mutational status as input for these methods. We concentrated on the current intriguing challenge: model comprehension and ease of use. Our comparison suggests that the tree-based methods are the most interpretable of those tested.

Management of aortic disease in children with FBN1-related Marfan syndrome.

Marfan syndrome (MFS) is a hereditary connective tissue disorder with an estimated prevalence of 1:5000-1:10 000 individuals. It is a pleiotropic disease characterized by specific ocular, cardiovascular, and skeletal features. The most common cardiovascular complication is aortic root dilatation which untreated can lead to life-threatening aortic root dissection, mainly occurring in adult patients. Prompt diagnosis, appropriate follow-up, and timely treatment can prevent aortic events. Currently there are no specific recommendations for treatment of children with MFS, and management is greatly based on adult guidelines. Furthermore, due to the scarcity of studies including children, there is a lack of uniform treatment across different centres. This consensus document aims at bridging these gaps of knowledge. This work is a joint collaboration between the paediatric subgroup of the European Network of Vascular Diseases (VASCERN, Heritable Thoracic Aortic Disease Working Group) and the Association for European Paediatric and Congenital Cardiology (AEPC). A group of experts from 12 different centres and 8 different countries participated in this effort. This document reviews four main subjects, namely, (i) imaging of the aorta at diagnosis and follow-up, (ii) recommendations on medical treatment, (iii) recommendations on surgical treatment, and (iv) recommendations on sport participation.

Effectiveness of catch-up and routine program of the 9-valent vaccine on cervical cancer risk reduction in Japan.

In 2013, the national human papillomavirus (HPV) immunization program began. However, in June 2013, Japan's Ministry of Health, Labor and Welfare (MHLW) announced a "temporary" suspension of its recommendation for the human papillomavirus vaccine. Finally, in November 2021, the MHLW ended its suspension of the recommendation of the HPV vaccine. To address the 9-year gap in HPV vaccinations the suspension had caused, the MHLW conducted a program of catch-up vaccinations from April 2022 to March 2025. Finally, in April 2023, the 9-valent HPV vaccine was approved for both the routine and catch-up vaccination programs in Japan. In this study, we investigated the potential effects of the introduction of the 9-valent vaccine on the increased risk of cervical cancer in females born after fiscal year (FY) 2000. We estimated the lifetime relative risk of cervical cancer incidence and death using the improved routine and catch-up vaccination rates after the recent resumption of the governmental recommendation for women and girls to have the HPV vaccination. These relative risks were calculated using a lifetime risk of 1.000 for cervical cancer incidence and death for females born in FY 1993. We predicted that even if a 90% vaccination rate were to be achieved by FY 2024 with the 9-valent vaccine among women born between FY 2000 and FY 2005, the risk would remain higher than for the vaccination generation. Therefore, for women born between FY 2000 and FY 2005, it will be necessary to significantly improve the cervical cancer screening rate to compensate for this increased risk.

Even though active recommendation for HPV vaccination has restarted, Japan's rates have not recovered.

Japan has a particularly critical situation surrounding its collapsed HPV vaccination program for preventing HPV-caused cervical cancers, a problem exacerbated by the lack of a national immunization database. We have determined the year-to-year HPV vaccination uptake by Japanese females and analyzed by birth fiscal year (FY) the monthly number of people receiving initial HPV vaccination. Our analysis covers the period from the start of public subsidies in 2010 to September 2023, using data provided by local governments. We calculated the cumulative number of monthly immunizations for those unimmunized as of April (the beginning of each vaccination year). The monthly number of initial HPV vaccinations was highest in August for every FY from FY 2010 to FY 2023; a second vaccination peak tended to occur in March when the vaccination year ended. The highest number of August vaccinations occurred in FY 2011, followed (in order) by 2012, 2021, 2022, 2023, and 2013. In Japan's ongoing catch-up vaccination program for young women, the monthly number of vaccinations increased in August 2022 but then slowed the following year. After FY 2021, the cumulative vaccination coverage of subjects unvaccinated at the beginning of the vaccination year but subsequently covered by routine immunizations was slightly improved. FY 2021 was when the governmental recommendations for HPV vaccination were resumed. More recent vaccination rates are considerably lower than those in FY 2011-2012 when vaccinations were first fully endorsed. Paralyzing HPV vaccination hesitancy, which began in FY 2013, will linger in Japan in FY 2024.

Low-Dose Continuous Kidney Replacement Therapy and Mortality in Critically Ill Patients With Acute Kidney Injury: A Retrospective Cohort Study.

RATIONALE & OBJECTIVE: Continuous kidney replacement therapy (CKRT) is preferred when available for hemodynamically unstable acute kidney injury (AKI) patients in the intensive care unit (ICU). The Kidney Disease: Improving Global Outcomes (KDIGO) guidelines recommend a delivered CKRT dose of 20-25mL/kg/h; however, in Japan the doses are typically below this recommendation due to government health insurance system restrictions. This study investigated the association between mortality and dose of CKRT. STUDY DESIGN: Single-center retrospective cohort study. SETTING & PARTICIPANTS: Critically ill patients with AKI treated with CKRT at a tertiary Japanese university hospital between January 1, 2012, and December 31, 2021. EXPOSURE: Delivered CKRT doses below or above the median. OUTCOME: 90-day mortality after CKRT initiation. ANALYTICAL APPROACH: Multivariable Cox regression analysis and Kaplan-Meier analysis. RESULTS: The study population consisted of 494 patients. The median age was 72 years, and 309 patients (62.6%) were men. Acute tubular injury was the leading cause of AKI, accounting for 81.8%. The median delivered CKRT dose was 13.2mL/kg/h. Among the study participants, 456 (92.3%) received delivered CKRT doses below 20mL/kg/h, and 204 (41.3%) died within 90 days after CKRT initiation. Multivariable Cox regression analysis revealed increased mortality in the below-median group (HR, 1.73 [95% CI, 1.19-2.51], P=0.004). Additionally, a significant, inverse, nonlinear association between 90-day mortality and delivered CKRT dose was observed using delivered CKRT dose as a continuous variable. LIMITATIONS: Single-center, retrospective, observational study. CONCLUSIONS: A lower delivered CKRT dose was independently associated with higher 90-day mortality among critically ill patients who mostly received dosing below the current KDIGO recommendations. PLAIN-LANGUAGE SUMMARY: The Kidney Disease: Improving Global Outcomes (KDIGO) guidelines recommend delivering a continuous kidney replacement therapy (CKRT) dose of 20-25mL/kg/h. However, it is not clear if it is safe to use delivered CKRT doses below this recommendation. In this study, over 90% of the patients received CKRT with a delivered dose below the KDIGO recommendation. We divided these patients into 2 groups based on the median delivered CKRT dose. Our findings show that a delivered CKRT dose below the median was associated with increased risk of death within 90 days. These findings show that a lower delivered CKRT dose was independently associated with higher 90-day mortality among critically ill patients who mostly received dosing below current KDIGO recommendations.

KDOQI Commentary on the KDIGO 2022 Update to the Clinical Practice Guideline for Diabetes Management in CKD.

The Kidney Disease: Improving Global Outcomes (KDIGO) guideline for diabetes management in chronic kidney disease (CKD) was updated in 2022, just 2 years after the previous update. The need for this rapid update is reflective of the recent and unprecedented positive results of numerous clinical trials aimed at reducing kidney and cardiovascular morbidity and mortality in people with diabetes. The Kidney Disease Outcomes Quality Initiative (KDOQI) work group for diabetes in CKD, convened by the National Kidney Foundation, provides herein a commentary on these changes, particularly the implications for health care in the United States. Changes to the KDIGO guideline mirror the evolution of sodium/glucose cotransporter 2 (SGLT2) inhibitors and glucagon-like peptide 1 receptor agonists from purely antihyperglycemic agents to cardiorenal-metabolic therapeutics, and the lower estimated glomerular filtration rate of≥20mL/min/1.73m2 for SGLT2 inhibitor initiation. New data have also brought the addition of the first-in-class, Federal Drug Administration-approved nonsteroidal mineralocorticoid receptor antagonist finerenone as an agent to reduce cardiorenal end points. While there has been significant progress in innovation, there remain serious challenges to implementation, particularly in the United States where inequities in insurance coverage and high costs limit their use, particularly in vulnerable populations, ultimately widening health care disparities.

Vessel wall MRI in giant cell arteritis: standardized protocol and scoring approach developed by an international working group.

OBJECTIVES: There are an increasing number of centers performing research on high-resolution vessel wall magnetic resonance imaging (VW-MRI) in giant cell arteritis (GCA). However, harmonized approaches to VW-MRI in GCA are lacking and are essential to performing multicentre studies. Using a data-driven, consensus-based approach, an international expert group developed a standardized MRI protocol and scoring system to advance multi-centered research in cranial GCA. METHODS: A targeted literature review of VW-MRI in cranial GCA was conducted. A working group comprised of radiologists, rheumatologists, and ophthalmologists with expertise in VW-MRI and GCA reviewed the results of the literature search, presented relevant data and images from their respective centers, and then reached consensus on recommendations related to key MRI structures, MRI sequences, scoring system, and other important considerations. RESULTS: A total of 21 relevant articles were identified and reviewed. Based on published literature, structures to be evaluated on MRI were categorized based on anatomic location (extradural cranial, intradural cranial, and orbits) and prioritization (core vs elective). Essential and elective sequences to comprehensively image cranial and orbital structures while minimizing scan time were determined along with scoring systems to grade contrast enhancement. CONCLUSION: This report describes a standardized approach to facilitate research of VW-MRI in cranial GCA that is the result of a multi-disciplinary, international collaboration of experts in VW-MRI and/or GCA.

Evidence of a bi-directional relationship between heart failure and diabetes: a strategy for the detection of glucose abnormalities and diabetes prevention in patients with heart failure.

Prevalence of heart failure (HF) and diabetes are markedly increasing globally. In a population of HF patients, approximately 40% have diabetes which is associated with a more severe HF, poorer cardiovascular outcomes and higher hospitalization rates for HF than HF patients without diabetes. Similar trends were shown in HF patients with prediabetes. In addition, the association between HF and renal function decline was demonstrated in patients with or without diabetes. However, the exact prevalence of dysglycemia in HF patients requires further investigation aiming to clarify the most accurate test to detect dysglycemia in this population. The relationship between HF and diabetes is complex and probably bidirectional. In one way, patients with diabetes have a more than two-fold risk of developing incident HF with reduced or preserved ejection fraction than those without diabetes. In the other way, patients with HF, when compared with those without HF, show an increased risk for the onset of diabetes due to several mechanisms including insulin resistance (IR), which makes HF emerging as a precursor for diabetes development. This article provides epidemiological evidence of undetected dysglycemia (prediabetes or diabetes) in HF patients and reviews the pathophysiological mechanisms which favor the development of IR and the risks associated with these disorders in HF patients. This review also offers a discussion of various strategies for the prevention of diabetes in HF patients, based first on fasting plasma glucose and HbA1c measurement and if normal on an oral glucose tolerance test as diagnostic tools for prediabetes and unknown diabetes that should be performed more extensively in those patients. It discusses the implementation of diabetes prevention measures and well-structured management programs for HF patients who are generally overweight or obese, as well as current pharmacotherapeutic options for prediabetes, including sodium-glucose cotransporter 2 inhibitors which are among the pillars of HF treatment and which recently showed a benefit in the reduction of incident diabetes in HF patients. Thus, there is an urgent need of routine screening for dysglycemia in all HF patients, which should contribute to reduce the incidence of diabetes and to treat earlier diabetes when already present.

Analysis of image data from the EuroNet PHL-C2 trial indicates a potential reduction in injected F-18 FDG activities in children: a proposal to update the EANM Paediatric Dosage Card.

BACKGROUND: The aim of this work is to provide the currently missing evidence that may allow an update of the Paediatric Dosage Card provided by the European Association of Nuclear Medicine (EANM) for conventional PET/CT systems. METHODS: In a total of 2082 consecutive [18F]FDG-PET scans performed within the EuroNet-PHL-C2 trial, the administered [18F]FDG activity was compared to the activity recommended by the EANM Paediatric Dosage Card. None of these scans had been rejected beforehand by the reference nuclear medicine panel of the trial because of poor image quality. For detailed quality assessment, a subset of 91 [18F]FDG-PET scans, all performed in different patients at staging, was selected according to pre-defined criteria, which (a) included only patients who had received substantially lower activities than those recommended by the EANM Paediatric Dosage Card, and (b) included as wide a range of different PET systems and imaging parameters as possible to ensure that the conclusions drawn in this work are as generally valid as possible. The image quality of the subset was evaluated visually by two independent readers using a quality scoring system as well as analytically based on a volume-of-interest analysis in 244 lesions and the healthy liver. Finally, recommendations for an update of the EANM Paediatric Dosage Card were derived based on the available data. RESULTS: The activity recommended by the EANM Paediatric Dosage Card was undercut by a median of 99.4 MBq in 1960 [18F]FDG-PET scans and exceeded by a median of 15.1 MBq in 119 scans. In the subset analysis (n = 91), all image data were visually classified as clinically useful. In addition, only a very weak correlation (r = 0.06) between activity reduction and tumour-to-background ratio was found. Due to the intended heterogeneity of the dataset, the noise could not be analysed statistically sound as the high range of different imaging variables resulted in very small subsets. Finally, a suggestion for an update of the EANM Paediatric Dosage Card was developed, based on the analysis presented, resulting in a mean activity reduction by 39%. CONCLUSION: The results of this work allow for a conservative update of the EANM Paediatric Dosage Card for [18F]FDG-PET/CT scans performed with conventional PET/CT systems.

Physiologically-based pharmacokinetic modelling guided dose evaluations of nirmatrelvir/ritonavir in renal impairment for the management of COVID-19.

We aimed to address factors contributing to the pharmacokinetic changes of nirmatrelvir/ritonavir in renal impaired (RI) patients and recommend dosing adjustment via a physiologically-based pharmacokinetic (PBPK) modelling approach. A PBPK model of nirmatrelvir/ritonavir was developed via Simcyp® Simulator. Sensitivity analysis of the influence of hepatic CYP3A4 intrinsic clearance and abundance, as well as hepatic non-CYP3A4 metabolism (other human liver microsomes [HLM] CLint) was performed to evaluate the effects of RI on oral clearance of nirmatrelvir. Other HLM CLint, the most sensitive parameter, was adjusted, and the simulated plasma concentration profiles of nirmatrelvir in severe RI subjects were within the therapeutic index of 292-10 000 ng/mL for dosing regimens of loading doses of 300/100 mg followed by 150/100 mg or 75/100 mg twice daily of nirmatrelvir/ritonavir. Considering that nirmatrelvir is available as a 150 mg tablet, we recommend 300/100 mg followed by 150/100 mg twice daily as the dosing regimen to be investigated in severe RI.

Young Age as a Predictor of Chemotherapy Recommendation and Treatment in Breast Cancer: A National Cancer Database Study.

INTRODUCTION: Breast cancer, although the second most common malignancy in women in the United States, is rare in patients under the age of 40 y. However, this young patient population has high recurrence and mortality rates, with chemotherapy frequently used as adjuvant treatment. We aimed to determine whether age is an independent predictor of chemotherapy recommendation and subsequent treatment and the relationship to Oncotype Dx (ODX) recurrence score (RS). METHODS: The National Cancer Database was retrospectively reviewed from 2010-2016 to identify women with early-stage (pT1-pT3, pN0-pN1mic, M0), hormone receptor positive, human epidermal growth factor receptor 2 negative breast cancer who underwent ODX RS testing. RESULTS: Of 95,382 patients who met the inclusion criteria, risk groups using the traditional ODX RS cutoffs were 59% low, 33% intermediate, and 8% high. Using Trial Assigning Individualized Options for Treatment RS cutoffs, risk groups were 23% low, 62% intermediate, and 15% high. Chemotherapy recommendation decreased as age at diagnosis increased (P < 0.001). Increasing age was associated with decreased odds of chemotherapy recommendation in univariate models both continuously (odds ratio: 0.98, 95% confidence interval 0.97-0.98; P < 0.001) and categorically by decade (P < 0.001). Age by decade remained an independent prognosticator of chemotherapy recommendation (P < 0.001), adjusted for risk groups. CONCLUSIONS: Chemotherapy recommendation and treatment differs by age among patients with early-stage hormone receptor positive breast cancer who undergo ODX testing. While molecular profiling has been shown to accurately predict the benefit of chemotherapy, younger age at diagnosis is a risk factor for discordant use of ODX RS for treatment strategies in breast cancer; with patients aged 18-39 disproportionately affected.

RSV vaccination strategies for high-risk patients 2023: a collaborative position paper by leading German medical societies and organizations.

Respiratory syncytial virus (RSV) inflicts severe illness and courses of infections not only in neonates, infants, and young children, but also causes significant morbidity and mortality in older adults and in people with immunosuppression, hemato-oncologic disease, chronic lung disease, or cardiovascular disease. In June and August 2023, effective vaccines against RSV were approved for the first time by the European Medicines Agency (EMA) for the EU. The respective pivotal studies showed a very high efficacy of the vaccine in preventing severe RSV-associated respiratory infections. At this point, use of the respective vaccines is restricted to persons aged 60 years or older, according to the registration studies. We therefore recommend use of the vaccination in persons aged 60 years or older. In addition, we recommend use of the vaccination in adults of any age with severe pulmonary or cardiovascular pre-existing conditions, as well as in adults with significant immune compromise, after individual consultation with the treating physician. Cost coverage can be applied for individually with the responsible health insurance company.

Treatment for central centrifugal cicatricial alopecia-Delphi consensus recommendations.

BACKGROUND: There is no established standard of care for treating central centrifugal cicatricial alopecia (CCCA), and treatment approaches vary widely. OBJECTIVE: To develop consensus statements regarding the use of various pharmacological therapies in treating adults with CCCA. METHODS: We invited 27 dermatologists with expertise in hair and scalp disorders to participate in a 3-round modified Delphi study between January and March 2023. Statements met strong consensus if 75% of respondents agreed or disagreed. Statements met moderate consensus if 55% or more but less than 75% agreed or disagreed. RESULTS: In round 1, 5 of 33 (15.2%) statements met strong consensus, followed by 9 of 28 (32.1%) in round 2. After the final round 3 meeting, strong consensus was reached for 20 of 70 (28.6%) overall statements. Two statements achieved moderate consensus. LIMITATIONS: This study included only English-speaking, US-based dermatologists and did not consider nonpharmacological therapies. CONCLUSION: Despite varying opinions among dermatologists, consensus was reached for several statements to help clinicians manage CCCA. We also highlight areas that lack expert consensus with the goal of advancing research and therapeutic options for CCCA.

Vaccination recommendations for adults receiving biologics and oral therapies for psoriasis and psoriatic arthritis: Delphi consensus from the medical board of the National Psoriasis Foundation.

BACKGROUND: For psoriatic patients who need to receive nonlive or live vaccines, evidence-based recommendations are needed regarding whether to pause or continue systemic therapies for psoriasis and/or psoriatic arthritis. OBJECTIVE: To evaluate literature regarding vaccine efficacy and safety and to generate consensus-based recommendations for adults receiving systemic therapies for psoriasis and/or psoriatic arthritis receiving nonlive or live vaccines. METHODS: Using a modified Delphi process, 22 consensus statements were developed by the National Psoriasis Foundation Medical Board and COVID-19 Task Force, and infectious disease experts. RESULTS: Key recommendations include continuing most oral and biologic therapies without modification for patients receiving nonlive vaccines; consider interruption of methotrexate for nonlive vaccines. For patients receiving live vaccines, discontinue most oral and biologic medications before and after administration of live vaccine. Specific recommendations include discontinuing most biologic therapies, except for abatacept, for 2-3 half-lives before live vaccine administration and deferring next dose 2-4 weeks after live vaccination. LIMITATIONS: Studies regarding infection rates after vaccination are lacking. CONCLUSION: Interruption of antipsoriatic oral and biologic therapies is generally not necessary for patients receiving nonlive vaccines. Temporary interruption of oral and biologic therapies before and after administration of live vaccines is recommended in most cases.

A patient information mining network for drug recommendation.

As an important task of natural language processing, medication recommendation aims to recommend medication combinations according to the electronic health record, which can also be regarded as a multi-label classification task. But patients often have multiple diseases simultaneously, and the model must consider drug-drug interactions (DDI) of medication combinations when recommending medications, making medication recommendation more difficult. There is little existing work to explore the changes in patient conditions. However, these changes may point to future trends in patient conditions that are critical for reducing DDI rates in recommended drug combinations. In this paper, we proposed the Patient Information Mining Network (PIMNet), which models the current core medications of patient by mining the temporal and spatial changes of patient medication order and patient condition vector, and allocates some auxiliary medications as the currently recommended medication combination. The experimental results show that the proposed model greatly reduces the recommended DDI of medications while achieving results no lower than the state-of-the-art results.

World Health Organization (WHO) guideline on the complementary feeding of infants and young children aged 6-23 months 2023: A multisociety response.

The recent World Health Organization (WHO) guideline aims to provide evidence-based recommendations on complementary feeding (CF) of healthy term infants and young children 6-23 months living in low-, middle-, and high-income countries, including both breastfed and non-breastfed children. Like WHO, our organizations aim to promote optimal infant and young child nutrition and health, with a focus on promoting breastfeeding as well as appropriate and timely CF. In this paper, we share our concerns about aspects of the guideline, some of which may have the potential to cause unintended harm in infants and young children and suggest alternative or modified proposals.

ACDNet: Attention-guided Collaborative Decision Network for effective medication recommendation.

Medication recommendation using Electronic Health Records (EHR) is challenging due to complex medical data. Current approaches extract longitudinal information from patient EHR to personalize recommendations. However, existing models often lack sufficient patient representation and overlook the importance of considering the similarity between a patient's medication records and specific medicines. Therefore, an Attention-guided Collaborative Decision Network (ACDNet) for medication recommendation is proposed in this paper. Specifically, ACDNet utilizes attention mechanism and Transformer to effectively capture patient health conditions and medication records by modeling their historical visits at both global and local levels. ACDNet also employs a collaborative decision framework, utilizing the similarity between medication records and medicine representation to facilitate the recommendation process. The experimental results on two extensive medical datasets, MIMIC-III and MIMIC-IV, clearly demonstrate that ACDNet outperforms state-of-the-art models in terms of Jaccard, PR-AUC, and F1 score, reaffirming its superiority. Moreover, the ablation experiments provide solid evidence of the effectiveness of each module in ACDNet, validating their contribution to the overall performance. Furthermore, a detailed case study reinforces the effectiveness of ACDNet in medication recommendation based on EHR data, showcasing its practical value in real-world healthcare scenarios.

Strategies for secondary use of real-world clinical and administrative data for outcome ascertainment in pragmatic clinical trials.

BACKGROUND: Pragmatic trials are gaining popularity as a cost-effective way to examine treatment effectiveness and generate timely comparative evidence. Incorporating supplementary real-world data is recommended for robust outcome monitoring. However, detailed operational guidelines are needed to inform effective use and integration of heterogeneous databases. OBJECTIVE: Lessons learned from the Veterans Affairs (VA) Diuretic Comparison Project (DCP) are reviewed, providing adaptable recommendations to capture clinical outcomes from real-world data. METHODS: Non-cancer deaths and major cardiovascular (CV) outcomes were determined using VA, Medicare, and National Death Index (NDI) data. Multiple ascertainment strategies were applied, including claims-based algorithms, natural language processing, and systematic chart review. RESULTS: During a mean follow-up of 2.4 (SD = 1.4) years, 907 CV events were identified within the VA healthcare system. Slight delays (∼1 year) were expected in obtaining Medicare data. An additional 298 patients were found having a CV event outside of the VA in 2016 - 2021, increasing the CV event rate from 3.5 % to 5.7 % (770 of 13,523 randomized). NDI data required âˆ¼2 years waiting period. Such inclusion did not increase the number of deaths identified (all 894 deaths were captured by VA data) but enhanced the accuracy in determining cause of death. CONCLUSION: Our experience supports the recommendation of integrating multiple data sources to improve clinical outcome ascertainment. While this approach is promising, hierarchical data aggregation is required when facing different acquisition timelines, information availability/completeness, coding practice, and system configurations. It may not be feasible to implement comparable applications and solutions to studies conducted under different constraints and practice. The recommendations provide guidance and possible action plans for researchers who are interested in applying cross-source data to ascertain all study outcomes.