Clinical Research Directory

A Prospective Validation Study of Radiomics in the Differential Diagnosis of Uterine Leiomyoma and Uterine Sarcoma

In our previous study, based on the multi-center clinical big data collected from January 2012 to January 2025, we have completed the construction of a multimodal early warning model for the malignant transformation of uterine fibroids. The model was mainly based on T2WI and DWI sequences, and was trained and optimized by support vector machine (SVM) algorithm. In the retrospective study and internal validation, the model shows high sensitivity and specificity, which preliminarily proves that it has good application potential in identifying high-risk groups and predicting the risk of malignant transformation of uterine fibroids. However, there are still some limitations in retrospective studies and internal validation results, and its application value, universality and stability in real clinical environment have not been fully verified. Therefore, we plan to conduct a prospective validation study in consecutive patients enrolled after January 2025 to evaluate the clinical performance and generalization of the model in predicting the malignant tendency or risk of malignant transformation of uterine fibroids through practical application in the real population, and further analyze the operability in the actual diagnosis and treatment process and the potential value for patient management. This study will provide reliable evidence for early screening, follow-up management and individualized treatment of high-risk population, and has important clinical and public health significance for improving the early diagnosis rate, reducing the risk of malignant transformation and improving the prognosis of patients with uterine fibroids.

Radiomics-Based Non-Invasive MRI Differentiation of Uterine Sarcomas and Fibroids

This retrospective case-control study aims to develop and validate a diagnostic model based on multimodal big data and artificial intelligence to differentiate uterine leiomyoma from uterine sarcoma. Investigators will extract historical case data from existing inpatient and outpatient records, including medical history, physical and gynecological examination findings, MRI imaging data, laboratory results, and pathological records. The study seeks to address the question of whether integrating diverse retrospective clinical data with advanced AI techniques can accurately classify uterine tumors as benign leiomyomas or malignant sarcomas, thereby supporting clinical decision-making and optimizing diagnostic workflows.

Artificial Intelligence Assisted Teaching of Structured Nursing Care According to ERAS Protocol

This study evaluates an AI-supported case-based educational approach for teaching nursing care of gynecologic oncology patients according to the Enhanced Recovery After Surgery (ERAS) protocol. This single-center, parallel-group randomized controlled trial was conducted during the second semester of the 2024-2025 academic year with third-year undergraduate nursing students who had successfully completed the Obstetrics and Gynecology Nursing course and its related clinical practice at a public university in western Türkiye. A total of 76 students were randomized in equal proportions to the AI-supported case-simulation group or the control group using block randomization. Six participants were lost to follow-up because of withdrawal or incomplete data collection forms, and the final analyses included 70 students (35 in each group). Data were collected before and after the intervention using the researcher-developed ERAS Protocol Nursing Care Knowledge Test and the Multidimensional 21st Century Skills Scale. Descriptive statistics, including frequencies, percentages, means, and standard deviations, were used to summarize participant characteristics and study variables. Baseline characteristics were examined using independent-samples t-tests for continuous variables and chi-square tests for categorical variables. Intervention effects were evaluated using 2 × 2 mixed-design analysis of variance (ANOVA), with time (pretest and posttest) as the within-subjects factor and group (AI-supported intervention and control) as the between-subjects factor. The principal effect of interest was the Time × Group interaction, indicating whether changes from pretest to posttest differed between groups. The same analytical approach was applied to ERAS knowledge, the total 21st-century skills score, and its subdimensions. Effect sizes were reported using partial eta squared (partial η²), and statistical significance was set at p \< .05.

AI Assistance for Detecting Active Bleeding During DSA Procedures

Clinically, DSA-VDet can continuously indicate suspected bleeding regions throughout a complete angiographic sequence and provide a second-reader opinion to the operator. It is expected to reduce the time required for frame-by-frame review and localization of culprit vessels, reduce missed detection of small, multiple, or unusually located bleeding sites, and improve consistency among physicians with different levels of experience. Faster culprit-vessel localization may also reduce repeat angiography, radiation exposure, and unnecessary superselective catheterization. These potential benefits will be tested in the prospective randomized trial; the study does not assume that the model will necessarily improve clinical outcomes. From a translational perspective, the study will establish a standardized temporal DSA database spanning multiple centers, devices, and anatomical regions, together with bleeding-site annotation standards, a model suitable for real-time deployment, and a pathway for clinical evaluation. These resources may provide a reproducible technical foundation for regional collaboration in emergency interventional care and AI-assisted interpretation in lower-resource hospitals. The core value of this study is therefore to address the 'single-region, weak-temporal, algorithm-centered, clinically under-evaluated' limitations of existing models. Using multicenter DSA videos from multiple anatomical regions, the study will leverage contrast kinetics and motion information to improve detection of small bleeding sites and will progress from retrospective validation through controlled physician interpretation to a prospective randomized trial, providing a continuous evaluation from technical performance to clinical benefit.

AI-driven Personalized Exercise Feedback Program on Exercise Adherence in Traumatic Brain Injury

This study aims to develop and evaluate an AI-driven Personalized Exercise Feedback Program (AI-PEF) to enhance exercise adherence and health outcomes in mTBI patients. Methods: AI-PEF integrates the transtheoretical model and self-determination theory with machine learning algorithms to provide real-time, personalized feedback. A phased randomized controlled trial will be conducted: Phase I evaluates feasibility and acceptability through Delphi methods with expert consensus and patient feedback; Phase II validates preliminary outcomes with 30 participants in a 2-arm randomized trial; and Phase III assesses the program's impact on exercise adherence, exercise motivation, mHealth app usability, functional capacity, sleep quality, depressive symptoms, concussion symptoms, and quality of life with 90 participants in a 3-arm randomized trial. In Phase III, quantitative and qualitative data will be collected in parallel, with qualitative data collected through semi-structured interviews to explore participants' experiences and perceptions of the intervention.

OpenEvidence Safety and Comparative Efficacy of Four LLM's in Clinical Practice

OpenEvidence is an online tool that aggregates and synthesizes data from peer-reviewed medical studies, then producing a response to a user's questions using generative AI. While it is in use by a number of clinicians (including residents) today, there is little to no published data on whether the tool's outputs are accurate and whether this information appropriately informs clinical decision making. Similarly, a number of clinicians are turning to other large language models (LLM's) to assist in decision making when providing clinical care. While there have been a number of studies published on the accuracy of these LLM's responses to medical boards questions or clinical vignettes, there have been few studies to date examining their performance in a real world clinical setting, and even fewer comparing this performance. In this study, investigators have two goals: 1. To determine whether the use of the AI tool "OpenEvidence" leads to clinically appropriate decisions when utilized by family medicine, internal medicine, and psychiatry residents in the course of clinical practice. 2. To determine how the output of the OpenEvidence tool compares with three other commonly-used, publicly-available large language models (OpenAI's ChatGPT, Anthropic's Claude, and Google's Gemini) in answering common questions that residents have in the course of clinical practice. To accomplish study goal #1, investigators have enlisted residents in the above specialties to use the OpenEvidence tool in the course of clinical practice. In order to mitigate any safety risks, the residents will also use a typical reference tool for their question, which is referred to as the "Gold Standard" tool. These tools include PubMed and UpToDate. The residents will: 1. State their clinical question. 2. Query OpenEvidence, capturing their prompt and the OpenEvidence output for data analysis. All residents will undergo training in prompt engineering at the start of the study. 3. State their clinical conclusion based on the OpenEvidence data. 4. Query the Gold Standard Resource. 5. State their final clinical conclusion. 6. Answer a question on whether their clinical conclusion was modified by the Gold Standard reference. 7. Answer a question on whether they had any clinical safety concerns on the output from OpenEvidence. Attending physician Subject Matter Experts (SMEs) matched by specialty with at least 5 years of post-training clinical experience will then evaluate the residents' responses. 5 years was chosen based the book "Outliers" by Malcolm Gladwell, in which he asserts that 10,000 hours of focused practice is needed to achieve expertise in a field. SMEs will be asked to evaluate the residents' initial clinical questions and their conclusions based only on OpenEvidence. They will be asked to rate the clinical appropriateness of those conclusions on a scale of 1-10. For questions where the SME's rate the clinical appropriateness of the residents' conclusions poorly (\< 5/10), they will be asked to review the OpenEvidence output and answer an additional question as to whether the output was incorrect or the resident misinterpreted the output from the tool. To accomplish goal #2, the initial prompt entered by the residents into OpenEvidence will be copied by the research team into ChatGPT, Gemini, and Claude. The outputs from each tool (including OpenEvidence) will be surfaced to SMEs, who will be asked to rate each output based on accuracy, completeness, and bias. Likert scales will be used for these ratings. SMEs will also be asked an open-ended question to identify any patient safety issues from any of the outputs.

AI-support in Emergency Medicine

The purpose of this study is to investigate whether the use of generative artificial intelligence (AI) as a support tool can improve physicians' ability to establish accurate differential diagnoses in emergency care. The main questions it aims to answer are: * Is AI-assisted differential diagnosis more accurate than the differential diagnosis performed by physicians? * Does physicians' differential diagnosis change - in terms of the number of correct diagnoses - when they have access to AI-assisted differential diagnosis?

A Feasibility Study of AI-Assisted Physiotherapy for Oral Cancer Patients

This study aims to evaluate the feasibility, safety, and acceptability of a newly developed artificial intelligence (AI)-assisted physiotherapy system for oromandibular and neck-shoulder range of motion (ROM) in patients who have undergone treatment for oral cancer. In this single-group, prospective, non-randomized pilot study, recruited participants will receive 4 to 6 weeks of AI-assisted physiotherapy. Participants will undergo a comprehensive clinical evaluation at baseline and post-intervention. During the intervention period, the AI system will perform a daily automated assessment to dynamically generate and adjust personalized exercise programs. Participants will perform these prescribed programs 3 to 4 times daily. Pre- and post-intervention changes, along with key feasibility parameters, acceptability, and safety metrics, will be statistically analyzed to inform future definitive trials.

Feasibility and Preliminary Performance of an AI Prototype for Digital ROSE During EBUS-TBNA and Peripheral TBNA: a Prospective Pilot Study (AI-ROSE-FEAS)

Rapid On-Site Evaluation (ROSE) of cytological slides obtained during EBUS-TBNA improves diagnostic yield by providing real-time adequacy assessment and preliminary diagnostic orientation after each needle pass. In centers without a dedicated cytopathologist, ROSE is performed by a second interventional pulmonologist acting as a dedicated ROSE operator (interventional pulmonologist-performed ROSE, IP-ROSE), a model associated with good but variable diagnostic performance compared to cytopathologist-performed ROSE. This study evaluates the feasibility and preliminary diagnostic performance of an investigational artificial intelligence prototype for digital ROSE. The prototype, developed in-house by the Principal Investigator, analyzes microscopic images of Diff-Quik stained cytological slides acquired through a dedicated digital microscope, together with basic clinical data, via API calls to a multimodal AI model. It produces two outputs: sample adequacy (appropriate/not appropriate) and malignancy suspicion (benign/malignant), each with a confidence score. The AI output is recorded in the study database for research purposes only and is not shown to the operator in real time; it does not influence clinical decisions during the procedure. The study is a prospective, monocentric, observational pilot study enrolling 65 adult patients undergoing EBUS-TBNA or peripheral TBNA with IP-ROSE at a single interventional pulmonology unit. The primary statistical unit is the individual ROSE slide, with an expected 130 to 160 evaluable slides. Co-primary endpoints are: (1) technical feasibility of the AI prototype, defined as the proportion of slides with valid AI output within 90 seconds; and (2) AI accuracy for sample adequacy assessment compared to the definitive cytopathological diagnosis, with an expected 95% confidence interval precision of ±5.5%. Secondary endpoints include AI accuracy for malignancy suspicion, agreement between the AI prototype and the IP-ROSE operator, and AI output latency. The AI prototype is not a commercially approved or CE-marked medical device. It was developed internally by the Principal Investigator for research purposes and is evaluated exclusively within this study. Data from this pilot study will inform the design of a subsequent confirmatory non-inferiority trial, which will be the subject of separate registration and ethical approval.

AI-Based Communication Simulation and Human Library Narratives.

This research plan aims to establish an effectiveness assessment system for promoting clinical communication, empathy, and emotion regulation in pediatric nursing students through a combination of AI communication simulation and human library narrative. This study sets five specific objectives: 1. To develop and implement a pediatric nursing teaching model that integrates AI communication simulation and human library narrative. 2. To evaluate the effectiveness of this teaching model in improving nursing students' clinical communication skills. 3. To examine the effectiveness of this teaching intervention in enhancing nursing students' empathy. 4. To explore the impact of this teaching intervention on nursing students' emotion regulation abilities. 5. To understand students' learning experiences, feelings, changes in emotion regulation, and suggestions for improvement regarding this integrated teaching intervention.

AI-Based Wound Monitoring: Automated Wound Progression Assessment Via Marker-Free Image Sequence

This study aims to develop a low-cost, marker-free intelligent wound assessment system that can analyze wound photos taken with a standard smartphone. By comparing wound images over time, the system will generate a quantifiable Wound Progression Index (WPI) to provide objective feedback on whether a wound is improving, stable, or worsening. The long-term goal is to support early detection of wound deterioration and improve wound care in both clinical and home settings.

Development and Validation of a Deep Learning Model to Predict Endodontic Retreatment Difficulty From Periapical Radiographs

The aim of this study is to develop and evaluate an artificial intelligence-based model capable of analyzing periapical radiographs of maxillary and mandibular molars to predict the difficulty level of non-surgical root canal retreatment. By integrating deep learning techniques with routinely acquired periapical radiographs, this study aims to enhance diagnostic support, improve clinical decision-making, and facilitate appropriate case selection or referral in endodontic practice.

AI in Assessing Aesthetic Outcomes in Rhinoplasty

This study aims to thoroughly assess the predictive accuracy of artificial intelligence-based nasal outcome simulations by comparing AI-generated preoperative predictions with objective postoperative nasal morphology using digital image analysis. To assess accuracy of AI-image measurement compared with imageJ software

AI-assisted CT for Risk Stratification in Coronary Artery Disease (ACTION)

The goal of this observational study is to learn if AI-assisted cardiac CT imaging can improve cardiovascular risk stratification and prediction of future coronary events in an adult population undergoing clinically indicated cardiac CT. The main questions it aims to answer are: * Can AI-enhanced cardiac CT accurately assess cardiovascular risk in a real-world adult population? * How do CT-derived plaque characteristics correlate with clinical, biochemical, and lifestyle risk factors? Researchers will compare subgroups (e.g., patients with different risk profiles, biomarkers, or imaging findings, and a subset undergoing OCT imaging) to see if differences in imaging and clinical parameters are associated with cardiovascular risk and plaque vulnerability. Participants will: * Provide informed consent and medical history/demographic information * Undergo blood sampling for cardiovascular and metabolic biomarkers * Have a resting ECG performed * Complete a detailed lifestyle and health questionnaire * Receive a non-invasive cardiac CT scan interpreted by an expert * Potentially receive heart rate-lowering medication (e.g., metoprolol) if required for imaging quality * Be referred for further clinical evaluation if clinically indicated 

Measuring AI Reliance Among Intern Doctors in Palestine

This study aims to enroll intern doctors and have them sit one of three identical radiology exams. The only difference between them is an AI-assistant. The differences between these groups will be used to measure the extent of AI reliance among intern doctors in Palestine.

Exploring the Potential of Artificial Intelligence for Earlier Breast Cancer Detection: A Retrospective Multi-Reader Study Based on AI-Assisted Mammographic Interpretation (EARLIEST-AI)

EARLIEST-AI Study type: Two-phase, multi-reader, blinded retrospective observational study based on data from a single clinic. The primary objective of the study is to assess the sensitivity and specificity of radiologists and Artificial Intelligence (AI) in interpreting mammographic examinations for breast cancer detection in a scenario where no previous examinations are available, and to compare the diagnostic performance of radiologists with and without AI support. The secondary objectives of the study are to assess the independent diagnostic performance of Computer Aided Detection (CAD) software, including sensitivity and specificity in identifying histopathologically confirmed breast cancer cases; to assess inter-reader and intra-reader variability in interpretation with and without AI support; to assess the agreement between AI outputs and histopathological findings; and to assess the impact of mammogram technical parameters on AI performance. Time frame: Review of a subset of mammograms randomly selected from those performed between January 1, 2012 and December 31, 2024. Imaging findings were classified according to the Breast Imaging Reporting and Data System (BI-RADS). Inclusion criteria: 1. Female patients aged 30 years or older. 2. One or more mammograms performed between January 1, 2012 and December 31, 2024. 3. Meets one of the following criteria: 3.1 Histopathologically confirmed diagnosis of breast cancer within 6 months of mammogram, 3.2 or two consecutive mammograms with BI-RADS 1 or BI-RADS 2. Exclusion criteria: 1. Mammograms of poor quality or artifacts that do not allow for reliable assessment. 2. History of breast surgery or previous breast cancer treatment that has significantly altered breast morphology. 3. Data deficiencies, including: 3.1. lack of previous histopathological data 3.2. or lack of available follow-up data.

Leveraging Large Language Models (LLM) to Enhance Research Competency Among Undergraduate Nursing Students: A Novel Approach to Research Education

The goal of this mixed method interventional study is to develop and test the effectiveness of integrating ChatGPT into the nursing research course to improve research competency among third-year undergraduate nursing students. The main questions it aims to answer is: Will participants who undergo the LLM-integrated curriculum show an increase in research competency and attitudes compared to participants who did not undergo this curriculum. Researchers will compare a students assessment grades, as well as their research competency and attitude, measured via the Research Competence Scale (R-Comp) and Revised Attitudes Towards Research scale (R-ATR) respectively. Research will determine whether the LLM-integrated curriculum could improve students understanding and attitudes towards research.

AI-Driven Multimodal Imaging Integration for Diagnosis and Prognostication of Digestive System Diseases

The goal of this observational, retrospective and prospective study is to develop a noninvasive disease assessment system by leveraging artificial intelligence (AI) to comprehensively analyze multi-modal imaging features, including magnetic resonance enterography (MRE) and computed tomography enterography (CTE), for the diagnosis and prognostication of digestive diseases. To this end, the investigators retrospectively enrolled imaging, endoscopic, and clinical data from 21 centers across China to construct and iteratively optimize the AI model. The model's performance will be prospectively validated in two centers, and its accuracy in lesion localization will be verified through real-world deployment in endoscopy suites.

Improving Artificial Intelligence-derived Algorithms for Estimating Length and Weight in NEonateS and infanTs up to 6 Months of Age (NEST)

The NEST study is a prospective, observational research study designed to collect clinical measurements and image data to develop and evaluate artificial intelligence (AI)-derived algorithms for estimating anthropometric parameters in neonates and young infants. The study focuses on infants from birth up to 6 months of age and aims to assess the accuracy of AI-based estimations of length, weight, and head circumference using photographs and/or video recordings captured during routine clinical care. These AI-derived measurements will be compared against standard clinical measurements obtained by trained healthcare professionals in neonatal and infant care settings.

AI-based Physiotherapy Evaluation System for Range of Motion in Oral Cancer Patients

This study aims to evaluate the validity and reliability of a novel AI-based physiotherapy evaluation system for measuring oromandibular and neck-shoulder range of motion (ROM). Traditional ROM assessments rely on manual measurements, which may be influenced by rater experience and variability. The proposed AI system uses automated keypoint tracking to provide objective and standardized measurements. In this cross-sectional study, healthy adult participants will perform standardized ROM tasks. Measurements obtained from the AI system will be compared with those from two independent raters using conventional clinical tools. Repeated measurements will be conducted to assess intra-rater and inter-rater reliability. The agreement between the AI system and human raters will be evaluated to determine the system's clinical applicability.

Advanced Symptom Palliation Through Integrated Relief Engagement

Beacon is a digital platform that processes objective and subjective aggregated data provided by patients. Objective data is provided by standard wearables, while subjective data is provided by patient-reported outcome measures (PROMs), comprising written and vocal patient reporting. The ASPIRE.AI study is a prospective study evaluating the feasibility of clinicians' use of aggregated data that was provided by patients and analyzed through "Beacon", and its influence on advanced cancer patients' palliative symptoms management. Approximately 40 consecutive eligible ambulatory advanced cancer patients first attending the palliative unit in the Davidoff Center will be enrolled. The trial will continue for \~1 year, with each patient participating in this trial for a total of about 12 weeks. All participants will receive the intervention. The intervention comprises the palliative standard of care treatment along with the usage of the Beacon digital platform, which enables comprehensive data collection and aggregation regarding the patient's biopsychosocial status, and thus, the patient's symptom burden. Data collected and aggregated through Beacon includes Beacon data provided by the patients via wearables (smartwatch/sensors), smartphones, and written and recorded PROMs. Researchers will then evaluate physician engagement with the platform, Influence on treatment, and the physician user experience rating as well as patients' adherence, satisfaction with Beacon usage, and changes in patients' symptom burden and quality of life.

Scientific Validity Assessment and Optimization of AI-Generated A3/A4 Type Questions for the Chinese Medical Licensing Examination

This is a cross-sectional study that primarily employs quantitative analysis, supplemented by qualitative assessment. The research is conducted in two stages: Phase I consists of a model performance comparison experiment, and Phase II involves an item quality evaluation experiment. The entire study adheres to the principles of single-blinding, randomization, and standardization to ensure scientific rigor and reproducibility. The single-blind design is implemented during the "standardized testing" phase, where the system intersperses AI-generated items with those authored by human experts. Participants remain blinded to the source of each item (AI-generated vs. human-authored) throughout the testing and scoring processes, thereby ensuring the objectivity of the evaluation results.

Validation on Clinical Adaptability of the Foundation Model Specific to Neuroimaging Diagnosis

This clinic trial aims to investigate whether artificial intelligence (AI) diagnostic tools at neurological diseases diagnosis on brain CT/MRI can improve the work efficiency of specialized neuroimaging physicians, with a specific focus on its clinical value in distinguishing normal from abnormal findings, critical value identification, and neurological disease classification. Using pathological and/or discharge diagnoses of neurological diseases as the gold standard, an AI model will be trained on over 10,000 CT/MRI cases to achieve diagnostic performance comparable to that of neurological radiologists before being transformed and putted to use. Furthermore, clinical trials will be conducted in sub-studies (abnormal cases identification, critical value assessment, and neurological disease classification) to validate the clinical utility of AI and human-AI collaboration in the precise diagnosis of neurological disorders. The expected outcomes include reducing missed and misdiagnosis rates, enabling rapid screening of critical conditions, and achieving precise imaging-based diagnosis by using AI tools.

Evaluation of Dora Care for Supporting Fracture Liaison Services (FLS)

What is the study about? This study is testing "Dora", an AI-powered assistant that can make phone calls to patients, for use in the Fracture Liaison Service (FLS). The FLS is a clinic that helps prevent more bone fractures after an initial "fragility fracture" (a break that happens easily, usually due to osteoporosis). Why is this being done? FLS clinicians often have to spend a lot of time on routine phone calls for assessments and follow-ups. If Dora can safely and accurately collect patient information, it might save time for staff and still give patients a good experience. What will happen to patients in the study? Invitation and consent - Patients with a new fragility fracture who are eligible will be invited to take part after informed consent. Dora call - Patients will receive an automated phone call from Dora, at the start of their FLS pathway and at follow-up. At intake, Dora will ask about risk factors for bone problems (e.g., smoking, alcohol use, family fracture history). At follow-up, Dora will ask about medication use, side effects, falls, or new fractures. Clinician call - Soon after, patients will have their usual phone appointment with an FLS clinician, who asks similar questions. Surveys/interviews - Patients will be asked to complete a short questionnaire and take part in an optional interview to say how they felt about talking to Dora. What about clinicians? Clinicians involved in the FLS pathway will be asked to complete a short survey and to take part in an optional interview to understand how useful Dora's reports might be in their work. Who can take part? Patients - Age 50+, English-speaking, with a new fragility fracture, and able to use the phone. Clinicians - Those working in FLS or similar bone health services. How long will it take? Each patient might be involved for up to about 7 months. The whole study will take about a year.