Citations

Abstract Motivation Illumina DNA methylation arrays have evolved rapidly, expanding genomic coverage while introducing backward incompatibilities by removing many CpG sites present in earlier versions. These changes result in systematic missing values when integrating data across array generations and substantially limiting the reuse of legacy datasets. Results We developed a two-stage framework for imputing missing DNA methylation values. The procedure first imputes randomly missing values using standard imputation techniques and then addresses systematic missingness using multi-output machine learning models, including support vector regression, nearest-neighbor methods, random forest models, and deep neural networks. When evaluated on real datasets with up to fifty percent induced missingness, the proposed framework consistently outperformed conventional imputation approaches. It also accurately imputes the missing CpG sites between methylation arrays and reduced representation bisulfite sequencing data, enabling robust cross-platform data integration. Analyses of large brain tumor methylation datasets demonstrate that the method restores array-specific methylation patterns while preserving biological complexity. Importantly, imputing missing methylation sites significantly improves the performance of epigenetic age prediction models. Availability and implementation This tool is implemented in the Python package “ultra-impute,” freely available at https://github.com/liguowang/ultra-impute. A code snippet demonstrating the usage of the ultra-impute package is provided in a Jupyter Notebook (https://github.com/liguowang/ultra-impute/blob/master/doc/Tutorial.ipynb).

Abstract Background We aimed to describe the real-world durability of clinical and endoscopic outcomes of upadacitinib (UPA) in ulcerative colitis (UC). Methods We retrospectively analyzed an ongoing multi-center cohort of UC patients(pts) in the United States who were started on UPA for active disease. Primary outcome was clinical remission at 1yr assessed by partial Mayo score(pMS)(resolution of all UC related symptoms and physician global assessment). Secondary outcomes assessed at various timepoints, including after re-escalation, included clinical response(>50%reduction in symptoms utilizing the pMS), endoscopic response (Mayo endoscopic score≤1or absence of erosions/ulcerations) and histological remission(defined as normal or chronic inactive). Durable secondary outcomes achieved at both 6months and at 1yr and beyond were also assessed. Descriptive statistics were performed. Adverse events(AEs) investigated included cardiovascular events and herpes zoster (HZ). Results 416pts were included(Table 1) with median values for age at initiation of36.00yrs(range27-48), disease duration7.00yrs(range3-14)and follow-up 993days(IQR,672-1103), with206(49.52%)with follow-up at 1yr or beyond. Of these, 66.1%had extensive colitis. Induction duration was8wks in 86.5%and16wks in 4.8 %(20/416). Most received an induction dose of45mg(97.1%)and maintenance dose of30mg(85.8%). Among the evaluable population at1yr and beyond, 62.5%(45/80)were in clinical remission, 75.9%(22/29)in endoscopic response with58.6%(17/29) in remission and66.6%(20/30)in histological remission. At 6months, clinical remission was51.6%(42/93), endoscopic response was 75%(24/32) of which43.75%(14/32)were in remission and histologic remission was seen in 40.8%(20/49).Durable secondary outcomes of clinical remission were seen in 71.4%(30/42), endoscopic improvement in 45.8%(11/24), and histologic remission in35%(7/20). Re-escalation to 45mg occurred in 35pts, of which48.6%(17/35)were re-escalated due to incomplete induction clinical response with all achieving response(11 in remission). The rest(18/35)were re-escalated for clinical relapse after initial remission where72.2%(13/18)were able to recapture remission. After re-escalation,54.3%(19/35)continued on 45mg daily. AEs were reported in12.7%with HZ in 7pts(2.1%). One pt aged53yrs had an episode of TIA after 1yr on UPA at 15mg, without pre-existing cardiovascular risk factors and UPA use is ongoing at the same dose. There were no reported episodes of worsening of pre-existing cardiovascular disease or DVT/PE. Conclusion UPA is durably effective and safe in a real-world clinical setting with re-escalation to 45mg capturing response in a majority of the pts. Ongoing recruitment will provide further evidence on long term effectiveness and safety. Conflict of interest: Ms. Huang, Katherine: No conflict of interest Chandnani, Arun: No conflict of interest Samaan, Sami: None Tan, Tingyi: I have no conflict of interests to declare. Ramesh, Prajith Raj: None Conlon, Caroline: No conflict of interest Klein, Jeremy: None Reddy, Nikhil: No conflict of interest Ayoub, Malek: None Cervantes Limon, Gabriela: none Myers, Melissa: None Bucaram, Martha: No conflict of interest Lieto, Stephen: No conflict of interest Scalzo, Nicholas: No conflict of interest Cohen, Benjamin: Abbvie - consulting and speaking ALPCO - consulting J & J Innovative Medicine - consulting Takeda - consulting and speaking Emmes Biopharma Services LLC - DSMB Pfizer - consulting Mohamed, Islam: nothing to declare Ciorba, Matthew: Advisory Board: Janssen, AbbVie, Pfizer, Geneoscopy Consulting: Janssen, AbbVie, Pfizer, Geneoscopy Grant contracts: Janssen, Pfizer, Incyte Shukla, Richa: Speakers bureau - Abbvie and Lilly Wright, Michael: None Xu, An Tao: No conflict of interest Chowla, Navreet: No conflict of interest Ogholikhan, Sina: Abbvie Medical Director Kujawski, Michelle: Abbvie Scientific Director Dulaney, David: Speaker bureau - Regeneron Speaker bureau - Sanofi Speaker bureau - Takeda Speaker bureau - Eli Lilly Fenster, Marc: none Benson, Caroline: No conflict of interest Alhobayb, Tamara: No conflict of interest Berinstein, Jeffrey: No conflict of interest Shivashankar, Raina: Abbvie and BMS - speaker bureau Janssen and Celltrion - grant for IBD fellowship funding Pfizer - consultant Bishu, Shrinivas: None Patel, Anish: speaker: abbvie, J & J, Lilly, BMS, Pfizer, Phathom, Takeda consultant: abbvie Ungaro, Ryan: Personal Fees: AbbVie, Bristol Myers Squibb, Genentech, Lilly, Pfizer, Janssen, Takeda Johnson, Amanda: Research grant: Abbvie, Spyre Therapeutics Pekow, Joel R.: CVS Health - Consulting Abbvie, Eli Lilly, Pfizer, Johnson and Johnson - Stocks Yarur, Andres: Personal Fees: Consultant for Takeda, Pfizer, Roche, Merck, Abbvie, Eli Lilly. Bristol Myers Squibb, Celltrion, Johnson and Johnson. Deepak, Parakkal: Parakkal Deepak has received research support under a sponsored research agreement unrelated to the data in the abstract from AbbVie, Johnson and Johnson, Sanofi, Merck, Teva, Direct Biologics, Tr1x, Boehringer Ingelheim, Bristol Myers Squibb, Pfizer, Prometheus Biosciences, Takeda Pharmaceuticals, Roche Genentech, Eli Lilly, AstraZeneca, Spyre and Agomab, has received consulting fees from Johnson and Johnson, Abbvie, Merck, Sobi, Celltrion, Fresenius Kabi, Asahi Kasei Pharma, Sandoz and CorEvitas, LLC and has served on the board of the Srategic Alliance for Intercultural Advocacy in GI.

Abstract Objectives To comprehensively evaluate the validity of International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes for both prevalent diagnoses and less common diseases, and to assess the performance of a large language model (LLM)-based system in validating these codes. Materials and Methods This retrospective study analyzed hospital admissions from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. We developed a validated LLM-based system using GPT-4o, refined through iterative prompt engineering, to assess ICD-10-CM code validity. We measured the positive predictive value (PPV) of ICD-10-CM codes, PPV of principal and secondary diagnoses, and the performance of an LLM-based system in code validation. Results Among 865 079 assigned codes, the PPV was 84.6% (95% CI, 84.5%-84.6%). Principal diagnoses had a PPV of 93.9% (95% CI, 93.7%-94.1%), while secondary diagnoses had a PPV of 83.8% (95% CI, 83.7%-83.9%). The LLM system demonstrated high performance in validating ICD codes, achieving 93.6% accuracy, 95.4% sensitivity, and 85.2% specificity. Among correctly assigned secondary diagnoses, the majority (67.9%) represented historical or baseline conditions, while 32.1% reflected active conditions that deviated from baseline status; 22.3% of these emerged after hospital admission. PPV decreases with later diagnosis positions, with the largest decline occurring between principal and secondary diagnoses. Discussion and Conclusion In this large-scale evaluation, ICD-10-CM codes exhibited generally high accuracy, though variability existed by position and condition type. A validated LLM system performed comparably to physician review and offers a scalable means to improve coding accuracy. These findings support the potential for integrating LLM-based auditing into routine workflows to strengthen the quality of administrative and research data.

Sharing research code in an open access version-controlled repository offers significant benefits for both science as a whole and for individual researchers. In this article, we focus on this practice, which is fully aligned with the NIH’s Gold Standard Science (GSS) program as well as FAIR (findable, accessible, interoperable, reusable) and TRUST (transparency, responsibility, user focus, sustainability, technology) principles. Gold Standard Science supports open science by emphasizing transparency, reproducibility, and the use of best practices that enable others to verify and extend research. Pairing a research article’s cited data snapshot with a versioned, environment-specific code release, deposited in a companion code repository, ensures that, upon submission to a medical journal, readers and reviewers can directly verify results. An executable and updatable companion code repository complements, rather than replaces, established research data repositories. When code underlying medical research results is made openly available, then other scientists can inspect, run, and validate analyses. These activities enhance reproducibility, which is a core aim of GSS. Shared code also facilitates collaborative innovation by allowing researchers to extend the utility of the code to new datasets and applications. For researchers, code sharing can increase visibility, credibility, and citation impact. Demonstrating transparency through shared executable and updatable code builds trust with journal readers, peer reviewers, funders, and peers. Shared code in an open access repository signals adherence to high standards of scientific integrity and attracts opportunities for collaboration. A researcher who shares code receives recognition as a leader in reproducible, trustworthy research consistent with NIH’s GSS principles.

ABSTRACT Objective To comprehensively evaluate the validity of ICD-10-CM codes for both prevalent diagnoses and less common diseases, and to assess the performance of a large language model (LLM)-based system in validating these codes. Materials and Methods This retrospective study analyzed hospital admissions from the Medical Information Mart for Intensive Care (MIMIC-IV) database. We developed a validated LLM-based system using GPT-4o, refined through iterative prompt engineering, to assess ICD-10-CM code validity. We measured the PPV of ICD-10-CM codes, PPV of principal and secondary diagnoses, and the performance of an LLM-based system in code validation. Results Among 865,079 assigned codes, the PPV was 84.6% (95% CI, 84.5%-84.6%). Principal diagnoses had a PPV of 93.9% (95% CI, 93.7%-94.1%), while secondary diagnoses had a PPV of 83.8% (95% CI, 83.7%-83.9%). The LLM system demonstrated high performance in validating ICD codes, achieving 93.6% accuracy, 95.4% sensitivity and 85.2% specificity. Among correctly assigned secondary diagnoses, the majority (67.9%) represented historical or baseline conditions, while 32.1% reflected active conditions that deviated from baseline status; 22.3% of these emerged after hospital admission. PPV decreases with later diagnosis positions, with the largest decline occurring between principal and secondary diagnoses. Discussion and Conclusion In this large-scale evaluation, ICD-10-CM codes exhibited generally high accuracy, though variability existed by position and condition type. A validated LLM system performed comparably to physician review and offers a scalable means to improve coding accuracy. These findings support the potential for integrating LLM-based auditing into routine workflows to strengthen the quality of administrative and research data.

Abstract Background Dynamically predicting patients at risk of Alzheimer's disease (AD) is crucial for timely treatment and care in precision medicine. Method We present a novel architecture, a Kolmogorov‐Arnold networks (KANs)‐based joint prediction model of longitudinal and survival data (JM‐KAN). Real‐life data of patients with Mild Cognitive Impairment (MCI) with no prior AD diagnosis from National Alzheimer's Coordinating Center (NACC) dataset was utilized for individualized predictions. We built joint models with landmark times every 1 year up to 4 years after the MCI diagnosis, and predicted survival outcomes one year after each landmark (i.e. 1, 2, 3, and 4 years). Integrated AUC (iAUC), and integrated Brier score (iBS) were measured for existing prediction models and our novel architecture. Result Out of 2711 unique patients with MCI, 821 patients progressed to develop AD. Mean time‐to‐event in years was 3.19 (standard deviation (SD): 2.29). The number of visits prior to AD diagnosis was 4.03 (SD: 2.41). With iAUC of 0.789, iBS of 0.118, JM‐KAN has both the highest iAUC and lowest iBS among joint models, which indicates superiority in discrimination and overall performance. For individualized predictions, the model successfully predicted both longitudinal risk factors and survival probability, as new information came in every year. Conclusion Using neural network‐based Artificial Intelligence (AI) model, we demonstrated that this architecture improves upon prior methods in dynamically updating predictions of AD as new longitudinal data becomes available.

Ultrahigh-field 7-T MRI offers superior resolution and contrast, which enable improved detection of subtle neuroanatomic abnormalities and disease-specific imaging biomarkers, with established clinical applications across a broad range of neurologic disorders.

Key Points The kidney microvasculature was quantified reliably in human autosomal dominant polycystic kidney disease early on, using super-resolution ultrasound localization microscopy. Microvascular parameters correlated with structural and functional indices, constituting promising biomarkers to evaluate noncystic parenchyma. Microvascular parameters may serve to assess disease severity and monitor progression from the early stages of the disease. Background In vitro studies in individuals with autosomal dominant polycystic kidney disease (ADPKD) have shown extensive intrarenal microvascular remodeling. However, analyses have been limited to the time of kidney failure because of the invasive nature of current methodologies to assess the intrarenal microvasculature. We hypothesized that super-resolution ultrasound microvascular imaging, known as ultrasound localization microscopy, enabled by high-framerate ultrasound imaging and microbubble localization and tracking postprocessing techniques, would safely, reliably, and reproducibly assess the intrarenal microvasculature in individuals with ADPKD at an early stage of the disease. Methods We prospectively enrolled young (<40 years of age), well-characterized, early-stage (eGFR >90 ml/min per 1.73 m 2 ) individuals with ADPKD, and sex–age 1:1 matched controls ( n =17 each) and assessed the intrarenal microvasculature using ultrasound localization microscopy. We determined the intersonographer reliability, intraobserver repeatability, and interobserver reproducibility of microvascular parameters in both groups, and the day-to-day variability in the control group. Microvascular parameters were correlated with structural (total kidney volume) and functional (eGFR) markers in individuals with ADPKD. Results Ultrasound localization microscopy provided a super-resolved map of the intrarenal microvasculature in ADPKD and matched controls. Intrarenal microvascular parameters ( e.g ., vessel density) were quantified in a safe, reliable (intraclass correlation coefficients, 0.96; 95% confidence interval, 0.91 to 0.98), reproducible (intraclass correlation coefficient, 0.96; 95% confidence interval, 0.93 to 0.98), and repeatable (coefficient of variations, 3.6%) manner. In control individuals, the day-to-day variability of most parameters was low, with coefficient of variation ranging from 1.8% to 10.7%. Microvascular parameters differed in individuals with ADPKD compared with controls, despite preserved eGFR. Interestingly, cortical vessel density and its derived perfusion correlated directly with eGFR, whereas tortuosity indices presented an inverse correlation. Conclusions Ultrasound localization microscopy provided a safe and reliable in vivo assessment of the human intrarenal microvasculature.