Wrangling Real-World Data: Optimizing Clinical Research Through Factor Selection with LASSO Regression.

To construct vocal recognition classification models using 6 machine learning algorithms and vocal emotional characteristics of individuals with subthreshold depression to facilitate early identification of subthreshold depression. We collected voice data from both normal individuals and participants with subthreshold depression by asking them to read specifically chosen words and texts. From each voice sample, 384-dimensional vocal emotional feature variables were extracted, including energy feature, Meir frequency cepstrum coefficient, zero cross rate feature, sound probability feature, fundamental frequency feature, difference feature. The Recursive Feature Elimination (RFE) method was employed to select voice feature variables. Classification models were then built using the machine learning algorithms Adaptive Boosting (AdaBoost), Random Forest (RF), Linear Discriminant Analysis (LDA), Logistic Regression (LR), Lasso Regression (LRLasso), and Support Vector Machine (SVM), and the performance of these models was evaluated. To assess generalization capability of the models, we used real-world speech data to evaluate the best speech recognition classification model. The AdaBoost, RF, and LDA models achieved high prediction accuracies of 100%, 100%, and 93.3% on word-reading speech test set, respectively. In the text-reading speech test set, the accuracies of the AdaBoost, RF, and LDA models were 90%, 80%, and 90%, respectively, while the accuracies of the other 3 models were all below 80%. On real-world word-reading and text-reading speech data, the classification models using AdaBoost and Random Forest still achieved high predictive accuracies (91.7% and 80.6% for AdaBoost and 86.1% and 77.8% for Random, respectively). Analyzing vocal emotional characteristics allows effective identification of individuals with subthreshold depression. The AdaBoost and RF models show excellent performance for classifying subthreshold depression individuals, and may thus potentially offer valuable assistance in the clinical and research settings.

Objectives Tinnitus is a phantom sound sensation without an external sound source. Due to its subjective and multifaceted nature it is measured using multi-item self-reported instruments. Many well-validated tinnitus-related questionnaires are available for clinical practice and scientific research, but so far no attention has been paid to their measurement invariance. The study aimed to examine measurement invariance of the Tinnitus Handicap Inventory with regard to gender and hearing impairment, and to identify the items that show differential item functioning (DIF) across the groups. Design This is a retrospective study using medical data from patients with tinnitus. They completed the Tinnitus Handicap Inventory (THI) and underwent pure-tone audiometry. Study sample 1106 adult patients with tinnitus (554 women and 552 men; 320 with normal hearing and 786 with hearing loss), aged 19–84 years. Results In the analysis, multi-group confirmatory factor analysis, hybrid ordinal logistic regression, Kernel smoothing in Item Response Theory, and lasso regression were applied. Measurement invariance was demonstrated across gender, but across hearing status the measurement was non-invariant. Five items were found to have DIF. Conclusions Researchers and clinicians should be aware of the potential risk of response bias when tinnitus severity is evaluated.

Predicting the price of a house helps for ascertain the house's selling price in a specific area and assist individuals in determining the ideal moment to purchase a home. Our goal in this machine learning task on house price prediction is to use data to develop a machine learning model capable of predicting housing values in the specified area. We will implement a linear regression algorithm on our dataset. By using real world data entities, we are going to predict the price of the house in that area. For better results we require data pre-processing units to increase the model's efficiency for this project we are using supervised learning, which is a part of machine learning. We have to go through different attributes of the dataset. This project provides us an overview on how to predict house prices using various machine learning models with the help of different python libraries. This suggested model is thought to be the most accurate one for estimating home prices and makes the most accurate predictions. This offers a succinct overview, which is necessary in order to forecast the price of the home. This project consists of what and how the house price model works with the assistance of machine learning technique using scikit-learn and which datasets we will be using in our proposed model. Key Words: house price, lasso regression, ridge regression, R-squared.

BackgroundArtificial intelligence has the potential to innovate current practices used to detect the imminent risk of suicide and to address shortcomings in traditional assessment methods.ObjectiveIn this paper, we sought to automatically classify short segments (40 milliseconds) of speech according to low versus imminent risk of suicide in a large number (n=281) of telephone calls made to 2 telehealth counselling services in Australia.MethodsA total of 281 help line telephone call recordings sourced from On The Line, Australia (n=266, 94.7%) and 000 Emergency services, Canberra (n=15, 5.3%) were included in this study. Imminent risk of suicide was coded for when callers affirmed intent, plan, and the availability of means; level of risk was assessed by the responding counsellor and reassessed by a team of clinical researchers using the Columbia Suicide Severity Rating Scale (=5/6). Low risk of suicide was coded for in an absence of intent, plan, and means and via Columbia suicide Severity Scale Ratings (=1/2). Preprocessing involved normalization and pre-emphasis of voice signals, while voice biometrics were extracted using the statistical language r. Candidate predictors were identified using Lasso regression. Each voice biomarker was assessed as a predictor of suicide risk using a generalized additive mixed effects model with splines to account for nonlinearity. Finally, a component-wise gradient boosting model was used to classify each call recording based on precoded suicide risk ratings.ResultsA total of 77 imminent-risk calls were compared with 204 low-risk calls. Moreover, 36 voice biomarkers were extracted from each speech frame. Caller sex was a significant moderating factor (β=–.84, 95% CI –0.85, –0.84; t=6.59, P<.001). Candidate biomarkers were reduced to 11 primary markers, with distinct models developed for men and women. Using leave-one-out cross-validation, ensuring that the speech frames of no single caller featured in both training and test data sets simultaneously, an area under the precision or recall curve of 0.985 was achieved (95% CI 0.97, 1.0). The gamboost classification model correctly classified 469,332/470,032 (99.85%) speech frames.ConclusionsThis study demonstrates an objective, efficient, and economical assessment of imminent suicide risk in an ecologically valid setting with potential applications to real-time assessment and response.Trial RegistrationAustralian New Zealand Clinical Trials Registry ACTRN12622000486729; https://www.anzctr.org.au/ACTRN12622000486729.aspx

Background: Cell-free DNA (cfDNA), shed by malignant tumor cells into extracellular fluid, provides valuable epigenetic information indicative of cancer status. Nipple aspirate fluid (NAF), a noninvasive liquid biopsy from at-risk women, contains nucleic acid and protein biomarkers from adjacent cancer cells, showing promise for breast cancer (BrC) detection. However, despite its potential, the application of cfDNA in NAF for BrC screening is still underexplored. This study aims to assess the utility and sensitivity of NAF cfDNA methylation signatures as biomarkers for early detection and risk stratification of BrC. Methods: A low-input cfDNA bisulfite sequencing (cfBS) protocol was developed and optimized for profiling genome-wide DNA methylation at single-base resolution using nanogram-scale cfDNA from NAF. A total of 76 NAF samples (17 BrC and 59 benign) were sequenced and analyzed. An epithelial-mesenchymal transition (EMT) methylation scoring metric was developed to quantify tumor phenotype. CpG methylation regions were identified using an in-house mean shift-based machine learning algorithm. Differentially methylated regions (DMRs) were identified using one-way ANOVA stratified by clinical stage. Predictive modeling was performed using random forest, LASSO regression, and support vector machine (SVM) classifiers. Results: When comparing cancer versus benign samples, differentially methylated regions were significantly enriched for EMT-related pathways. EMT scores were significantly elevated in BrC samples (p &amp;lt; 0.01) and positively correlated with clinical risk scores (R = 0.32, p = 0.0043). Validation in the Cancer Genome Atlas (TCGA-BRCA) cohort confirmed that methylation alterations in key EMT-related regions were strongly associated with corresponding changes in gene expression. One-way ANOVA identified 1,737 CpG regions with methylation levels significantly associated with increasing clinical risk. SVM-based predictive models using these methylation signatures achieved &amp;gt;90% accuracy in distinguishing BrC from benign cases, including early-stage tumors. These signatures consisted of 30 genes involved in the Gene Ontology biological process regulation of intracellular signal transduction. Conclusions: Our findings demonstrate that cfDNA methylation profiling from NAF is both technically feasible and biologically informative, offering a highly sensitive, noninvasive strategy for early BrC detection. By capturing EMT-associated epigenetic alterations, NAF cfBS has the potential to identify aggressive phenotypes at early stages. The high predictive accuracy achieved through machine learning models highlights its clinical utility as a complementary tool to existing screening modalities, especially for women with inconclusive mammograms, which is especially common in women with dense breast tissue. These results strongly support further validation in larger, prospective cohorts and suggest that NAF-based liquid biopsy could be integrated into personalized screening frameworks to improve early detection, reduce unnecessary biopsies, and guide preventative care. Financial Disclosures: The study is supported by the Texas A&amp;M University Presidential Clinical Research Partnership program. Disclaimer: Opinions expressed by the authors are their own, and this material should not be interpreted as representing the official viewpoint of the US Department of Health and Human Services, the National Institutes of Health, the National Cancer Institute, or the Division of Cancer Prevention. Citation Format: M. Jeon, E. Sauter, K. Zhang. Nipple Aspirate Cell-free DNA Methylation Signatures for Early Breast Cancer Detection [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS1-06-20.

In health research discourse, there is the assumption that a controlled, planned clinical research naturally produces cleaner datasets, while real-world data suffers from quality deficiencies due to its spontaneous collection without anticipation of secondary research use. However, this binary view simplifies a complex reality. Both clinical research and real-world data exhibit quality issues, challenging the convention that data origin determines data integrity. What proves to be more vital is the way data quality (DQ) is being assessed—particularly when examining the exhibited relationships between data items that reflect the true complexity of healthcare information. Health data is inherently interconnected e.g., height relates to weight and age; medication dosages connect to patient characteristics and medical conditions. Yet most data quality assessment (DQA) studies limit their logical evaluations to only a handful of interdependent items, missing the broader web of relationships that could reveal critical quality issues. Contradictions within datasets serve as an important DQ indicator (measurable attribute of data) that extend far beyond simple data entry errors. When values contradict each other, they reveal misunderstandings of domain knowledge, failures in data capture infrastructure, and problems in data transformation processes. Given the complexity of the contradiction patterns, a comprehensive assessment becomes essential. Comprehensive DQA in health research requires evaluating data on multiple dimensions, such as completeness, conformance, and plausibility, to determine the suitability of data for intended use. These dimensions are used to group related measurable attributes of data referred to as indicators. While some DQ indicators are implemented as generic rules (e.g., completeness), contradiction, which denotes pair(s) of measurements or facts that cannot logically coexist, are multi-dimensional. Some contradictions require common knowledge for their evaluation—e.g., the disparity in age information at baseline and during follow-up visits. However, contradictions in the pre-analytic states of blood samples require knowledge of experts. Although various tools have been developed to assess contradictions in health data, the evolving nature of contradiction rules presents the following challenges for existing tools: 1) limited reusability across diverse health datasets due to incompatible predefined rules, 2) insufficient contextual information required for comprehensive contradiction assessment, and 3) lack of consistent representation of contradiction patterns to support transparent rule implementations and computational efficiency. To address these gaps in a systematic manner, the first step of this thesis was to adapt an existing tool developed based on an existing framework to assess broader contradiction rules introduced in the German Corona Consensus (GECCO) dataset. Building on the experience from the modification of Schmidt et al.'s tool, this thesis developed a new framework that generalizes the DQA tool to accept custom contradiction rules compatible with diverse health datasets. This is intended to empower domain experts in defining effective contradiction rules on an ad-hoc basis while maintaining alignment with established DQ indicators. Although the primary dependencies between health data items form the basis for contradiction rule definition, the required context for a comprehensive analysis is equally relevant. For example, diabetes mellitus (DM) and insulin medication (INS) are two interdependent items, however, it is the timestamps that indicate if the INS was administered before the DM diagnosis, which is a prelude for a conclusive contradiction finding. Through a qualitative grading scheme for analyzing metadata defined in a study database, this work created a mechanism to measure the availability of the contextual information required to support a context-aware contradiction analysis. However, having access to rich contextual information is only part of the solution—the way contradiction rules are structured and implemented is equally critical for effective DQA. While researchers have worked extensively on harmonizing DQ indicators, the challenge extends beyond categorization to implementation efficiency. Though the taxonomies for contradictions, such as logical and empirical contradictions, reflect the domain description, the rules implemented in the DQA systems do not rely on semantics. Extracting the structure of the varied contradiction patterns ensures transparency and consistency in rule implementation. This work evaluated the performance of different rule implementations and offered an optimization method that integrates the fastest and traceable units of the rules in one unified implementation. This is relevant because broadly defined DQA rules can cause performance degradation in large database infrastructures. With these several perspectives on contradictions as a DQ indicator, the contributions in this thesis enhance contradiction assessment by improving reusability across diverse health datasets, enabling context-aware assessments, and optimizing rule implementation for transparency and computational efficiency. This systematic approach supports high-quality health data assessment while maintaining traceability to support data cleansing efforts.

This paper considers the problem of detecting and localizing material anomalies in solid structures, given spatiotemporal observations at a pre-defined grid of points that collectively describe the material displacement resulting from an induced, propagating acoustic surface wave. We propose an approach that seeks to separate or “demix” each temporal snapshot of the propagating wavefield into its constituent components, which are assumed to be morphologically dissimilar in the vicinity of material defects. We cast this demixing approach as a group lasso regression task, characterized by morphologically dissimilar dictionaries, and establish conditions under which material anomalies may be accurately identified using this approach. We demonstrate and validate the performance of this approach on synthetic data as well as real-world data.

BackgroundThe analysis of real-world data in clinical research is rising, but its use to study dementia subtypes has been hardly addressed. We hypothesized that real-world data might be a powerful tool to update AD epidemiology at a lower cost than face-to-face studies, to estimate the prevalence and incidence rates of AD in Catalonia (Southern Europe), and to assess the adequacy of real-world data routinely collected in primary care settings for epidemiological research on AD.MethodsWe obtained data from the System for the Development of Research in Primary Care (SIDIAP) database, which contains anonymized information of > 80% of the Catalan population. We estimated crude and standardized incidence rates and prevalences (95% confidence intervals (CI)) of AD in people aged at least 65 years living in Catalonia in 2016.ResultsAge- and sex-standardized prevalence and incidence rate of AD were 3.1% (95%CI 2.7–3.6) and 4.2 per 1000 person-years (95%CI 3.8–4.6), respectively. Prevalence and incidence were higher in women and in the oldest people.ConclusionsOur incidence and prevalence estimations were slightly lower than the recent face-to-face studies conducted in Spain and higher than other analyses of electronic health data from other European populations. Real-world data routinely collected in primary care settings could be a powerful tool to study the epidemiology of AD.

Real-world data (RWD) applications in healthcare that support learning health systems and pragmatic clinical trials are gaining momentum, largely due to legislation supporting real-world evidence (RWE) for drug approvals. Clinical notes are thought to be the cornerstone of RWD applications, particularly for conditions with limited effective treatments, extrapolation of treatments from other conditions, or heterogenous disease biology and clinical phenotypes. Here, we discuss current issues in applying RWD captured at the point-of-care and provide a framework for clinicians to engage in RWD collection. To achieve clinically meaningful results, RWD must be reliably captured using consistent terminology in the description of our patients. RWD complements traditional clinical trials and research by informing the generalizability of results, generating new hypotheses, and creating a large data network for scientific discovery. Effective clinician engagement in the development of RWD applications is necessary for continued progress in the field.

Real world data (RWD) and real-world evidence (RWE) plays an increasingly important role in clinical research since scientific knowledge is obtained during routine clinical large-scale practice and not experimentally as occurs in the highly controlled traditional clinical trials. Particularly, the electronic health records (EHRs) are a relevant source of data. Nevertheless, there are also significant challenges in the correct use and interpretation of EHRs data, such as bias, heterogeneity of the population, and missing or non-standardized data formats. Despite the RWD and RWE recognized difficulties, these are easily outweighed by the benefits of ensuring the efficacy, safety, and cost-effectiveness in complement to the gold standards of the randomized controlled trial (RCT), namely by providing a complete picture regarding factors and variables that can guide robust clinical decisions. Their relevance can be even further evident as healthcare units develop more accurate EHRs always in the respect for the privacy of patient data. This editorial is an overview of the RWD and RWE major aspects of the state of the art and supports the Special Issue on “Digital Health and Big Data Analytics: Implications of Real-World Evidence for Clinicians and Policymakers” aimed to explore all the potential and the utility of RWD and RWE in offering insights on diseases in a broad spectrum.

Large-scale real-world data (RWD) are increasingly used in clinical and epidemiological research, although database-specific structures and limitations may affect study validity and applicability. The Taiwan National Health Insurance Research Database (NHIRD) and the TriNetX network are two widely used RWD sources. This review compares their key features, strengths, and limitations and discusses approaches to address methodological challenges in real-world studies. The NHIRD comprises comprehensive, population-based, longitudinal claims data covering nearly the entire Taiwanese population. Its strengths include minimal selection bias and broad follow-up capacity. However, limitations include infrequent updates, limited clinical detail, and a Taiwan-specific context that may restrict generalizability. In contrast, TriNetX is a multinational federated network of electronic medical records from diverse healthcare systems, offering larger and more heterogeneous populations, richer clinical variables, and near real-time analytic capability, but with potential hospital-based selection bias and limited flexibility due to its fixed analytic interface. Representative studies published between 2010 and 2024 demonstrate the application of both databases across multiple medical disciplines. To mitigate data-related limitations, commonly used strategies include refined inclusion and exclusion criteria, proxy variables for unavailable measures, and triangulation with external datasets, which can strengthen study validity and interpretability. NHIRD and TriNetX are complementary real-world data sources, each with distinct strengths and limitations. Aligning research objectives with database characteristics is essential for appropriate study design. Recognition of platform-specific trade-offs and application of targeted methodological strategies support the validity and generalizability of real-world evidence.

Evidence provided by traditional clinical research often fails to answer patients’, physicians’ and healthcare decision-makers’ questions about real-world practice and outcomes; in fact, real-world effectiveness may considerably differ from efficacy assessed in clinical trial settings. The rise of interest in real-world data (RWD) - data routinely collected outside a controlled research environment - is driven by the increasing need for evidence in specific populations, such as comorbid or multi-treated people. RWD can also allow investigation of unanticipated, uncommon or long-term outcomes. In addition, RWD may represent the only source of information in some fields of special interest, e.g. rare diseases. Furthermore, current conditional reimbursement systems of drugs require dynamic evaluations of cost-effectiveness that have necessarily to take RWD into account. Indeed, RWD play a relevant role also in Health Technology Assessment (HTA) projects that should release the most up-to-date evidence and should be updated once new data are available. Nonetheless, common concerns about evidence derived from RWD include uncertainty about data quality, high possibility of bias, difficulties in managing a vast amount of data coming from different sources as well as legal and privacy requirements, in particular in the light of the new General Data Protection Regulation. However, HTA cannot exempt using RWD as it is entrusted to assess short- and long-term consequences of the application of health technologies, in terms of both health outcomes and costs. This workshop is an opportunity to discuss current and future challenges in the use of RWD in HTA and concerns preventing researchers from exploiting RWD full potential. Key messages Real-world evidence and randomized control trial data are considered mutually complementary in generating evidence and supporting decision-making. Real-world data have proved to be crucial as instrument to gather data in HTA and to support decision-making.

This study introduces a Lasso–Prophet hybrid framework developed to deal with the limitations and gap of Facebook’s Prophet model. The approach begins with Prophet’s decomposition of a time series into its fundamental components trend, seasonality, and holiday effects, and then applies Lasso regression to the residuals to capture additional structural patterns that fail to capture by the base model. This layered methodology boosts predictive accuracy by enabling the model to learn both systematic and irregular variations within temporal data by incorporating Lasso’s feature selection capability, the framework efficiently handles highdimensional datasets, retaining only the most informative predictors. The outcome is a hybrid model which achieves an optimal balance among interpretability, scalability, sparsity, and forecasting precision. Validation on simulated high-feature datasets and real-world electricity consumption data demonstrates that the Lasso–Prophet hybrid consistently outperforms the Prophet and other baseline models.OPEN ACCESS Received: 31/08/2025 Accepted: 17/11/2025 Published: 23/01/2026

Data-Driven Approaches to Maximize the Impact of Pediatric Clinical Trials.

The development of new treatments for gynecological malignancies has been conducted mainly through collaborative international phase III trials led by the United States and Europe. The survival outcomes of many gynecological malignancies have greatly improved as a result. Recent large-scale genome-wide association studies have revealed that drug efficacy and adverse event profiles are not always uniform. Thus, it is important to validate new treatment options in each country to safely and efficiently provide newly developed treatment options to patients with gynecological malignancies. The Japanese Gynecologic Oncology Group (JGOG) is conducting 5 cohort studies (JGOG 3026, 3027, 3028, 3030, and 3031) to establish real-world data (RWD) of poly(ADP-ribose) polymerase (PARP) inhibitor use in patients with advanced or recurrent epithelial ovarian cancer. The RWD constructed will be used to provide newly developed PARP inhibitors for women with advanced or recurrent ovarian cancer in a safer and more efficient manner as well as to develop further treatment options. In 2022, The JGOG, Korean Gynecologic Oncology Group, Chinese Gynecologic Cancer Society, and Taiwanese Gynecologic Oncology Group established the East Asian Gynecologic Oncology Trial Group to collaborate with East Asian countries in clinical research on gynecologic malignancies and disseminate new knowledge on gynecologic malignancies from Asia. The JGOG will conduct a collaborative integrated analysis of the RWD generated from Asian countries and disseminate real-world clinical knowledge regarding new treatment options that have been clinically implemented.