Bias in medical AI: Implications for clinical decision-making

PLOS Digital Health, Год журнала: 2024, Номер 3(11), С. e0000651 - e0000651

Опубликована: Ноя. 7, 2024

Biases in medical artificial intelligence (AI) arise and compound throughout the AI lifecycle. These biases can have significant clinical consequences, especially applications that involve decision-making. Left unaddressed, biased lead to substandard decisions perpetuation exacerbation of longstanding healthcare disparities. We discuss potential at different stages development pipeline how they affect algorithms Bias occur data features labels, model evaluation, deployment, publication. Insufficient sample sizes for certain patient groups result suboptimal performance, algorithm underestimation, clinically unmeaningful predictions. Missing findings also produce behavior, including capturable but nonrandomly missing data, such as diagnosis codes, is not usually or easily captured, social determinants health. Expertly annotated labels used train supervised learning models may reflect implicit cognitive care practices. Overreliance on performance metrics during obscure bias diminish a model's utility. When applied outside training cohort, deteriorate from previous validation do so differentially across subgroups. How end users interact with deployed solutions introduce bias. Finally, where are developed published, by whom, impacts trajectories priorities future development. Solutions mitigate must be implemented care, which include collection large diverse sets, statistical debiasing methods, thorough emphasis interpretability, standardized reporting transparency requirements. Prior real-world implementation settings, rigorous through trials critical demonstrate unbiased application. Addressing crucial ensuring all patients benefit equitably AI.

Язык: Английский

---

Advancements in Machine Learning and Artificial Intelligence in the Radiological Detection of Pulmonary Embolism

Cureus, Год журнала: 2025, Номер unknown

Опубликована: Янв. 29, 2025

Pulmonary embolism (PE) is a clinically challenging diagnosis that varies from silent to life-threatening symptoms. Timely of the condition subject clinical assessment, D-dimer testing and radiological imaging. Computed tomography pulmonary angiogram (CTPA) considered gold standard imaging modality, although some cases can be missed due reader dependency, resulting in adverse patient outcomes. Hence, it crucial implement faster precise diagnostic strategies help clinicians diagnose treat PE patients promptly mitigate morbidity mortality. Machine learning (ML) artificial intelligence (AI) are newly emerging tools medical field, including imaging, potentially improving efficacy. Our review studies showed computer-aided design (CAD) AI displayed similar superior sensitivity specificity identifying on CTPA as compared radiologists. Several demonstrated potential minor scans showing promising ability aid reducing substantially. However, imperative sophisticated conduct large trials integrate use everyday setting establish guidelines for its ethical applicability. ML also physicians formulating individualized management enhance

Язык: Английский

---

AI-Driven Evolution in Teledentistry: A Comprehensive Overview of Technology and Clinical Applications

Dentistry Review, Год журнала: 2025, Номер unknown, С. 100154 - 100154

Опубликована: Март 1, 2025

Язык: Английский

---

Principles for enhancing trust in artificial intelligence systems among medical imaging professionals in Ghana: A nationwide cross-sectional study

Radiography, Год журнала: 2025, Номер 31(3), С. 102953 - 102953

Опубликована: Апрель 14, 2025

Язык: Английский

---

A Comprehensive Review on the Application of Artificial Intelligence for Predicting Postsurgical Recurrence Risk in Early‐Stage Non‐Small Cell Lung Cancer Using Computed Tomography, Positron Emission Tomography, and Clinical Data

Journal of Medical Radiation Sciences, Год журнала: 2025, Номер unknown

Опубликована: Янв. 23, 2025

ABSTRACT Introduction Non‐small cell lung cancer (NSCLC) is the leading cause of cancer‐related mortality worldwide. Despite advancements in early detection and treatment, postsurgical recurrence remains a significant challenge, occurring 30%–55% patients within 5 years after surgery. This review analysed existing studies on utilisation artificial intelligence (AI), incorporating CT, PET, clinical data, for predicting risk early‐stage NSCLCs. Methods A literature search was conducted across multiple databases, focusing published between 2018 2024 that employed radiomics, machine learning, deep learning based preoperative positron emission tomography (PET), computed (CT), PET/CT, with or without data integration. Sixteen met inclusion criteria were assessed methodological quality using METhodological RadiomICs Score (METRICS). Results The reviewed demonstrated potential radiomics AI models postoperative risk. Various approaches showed promising results, including handcrafted features, models, multimodal combining different imaging modalities data. However, several challenges limitations identified, such as small sample sizes, lack external validation, interpretability issues, need effective techniques. Conclusions Future research should focus conducting larger, prospective, multicentre studies, improving integration interpretability, enhancing fusion modalities, assessing utility, standardising methodologies, fostering collaboration among researchers institutions. Addressing these aspects will advance development robust generalizable NSCLC, ultimately patient care outcomes.

Язык: Английский

---

Artificial intelligence-enabled obesity prediction: A systematic review of cohort data analysis

International Journal of Medical Informatics, Год журнала: 2025, Номер 196, С. 105804 - 105804

Опубликована: Янв. 25, 2025

Язык: Английский

---

The Future of Virology Diagnostics Using Wearable Devices Driven by Artificial Intelligence

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 473 - 504

Опубликована: Янв. 10, 2025

The utilization of the wearable devices (WDs) that are enhanced by artificial intelligence (AI) can have a notable potential in healthcare. This chapter aimed to provide an overview applications AI-driven WDs enhancing early detection and management virus infections. First, we presented examples highlight capabilities very monitoring infections such as COVID-19. In addition, provided on utility machine learning algorithms analyze large data for signs We also overviewed enable real-time surveillance effective outbreak management. showed how this be achieved via collection analysis diverse WDs' across various populations. Finally, discussed challenges ethical issues comes with virology diagnostics, including concerns about privacy security well issue equitable access.

Язык: Английский

---

DETECT IPN: Real-World Experience with Automated Detection of Incidental Pulmonary Nodules in an All-Comer Population

Open Journal of Radiology, Год журнала: 2025, Номер 15(01), С. 13 - 25

Опубликована: Янв. 1, 2025

Язык: Английский

---

Artificial intelligence bias auditing – current approaches, challenges and lessons from practice

Review of Accounting and Finance, Год журнала: 2025, Номер unknown

Опубликована: Март 10, 2025

Purpose This study aims to explore current approaches, challenges and practical lessons in auditing artificial intelligence (AI) systems for bias, focusing on legal compliance audits the USA European Union (EU). emphasizes need standardized methodologies ensure trustworthy AI that align with ethical regulatory expectations. Design/methodology/approach A qualitative analysis compared bias audit practices, including US report summaries under New York City’s Local Law 144 conformity assessments (CAs) required by EU Act. Data was gathered from publicly available reports guidelines identify key lessons. Findings The findings revealed are susceptible various biases stemming data, algorithms human oversight. Although valuable, lack standardization, leading inconsistent reporting practices. EU’s risk-based CA approach offers a comprehensive framework; however, its effectiveness depends developing standards consistent application. Research limitations/implications is limited early implementation stage of frameworks, particularly Act, restricted access reports. geographic focus jurisdictions may limit generalizability findings. availability constraints frameworks affect comparative analysis. Future research should longitudinal studies effectiveness, development intersectional assessment investigation automated tools can adapt emerging technologies while maintaining feasibility across different organizational contexts. Practical implications underscores necessity adopting socio-technical perspectives auditing. It provides actionable insights firms, regulators auditors into implementing robust governance risk practices mitigate biases. Social Effective algorithmic fairness prevent discriminatory outcomes critical domains like employment, health care financial services. emphasize enhanced stakeholder engagement community representation processes. Implementing help close socioeconomic gaps identifying mitigating disproportionately affecting marginalized groups. contributes equitable respect diversity promote social justice technological advancement. Originality/value discourse comparing two CAs implementation. highlights role standardization advancing finance accounting

Язык: Английский

---

Machine-learning-based diagnosis and progression analysis of knee osteoarthritis

Discover Data, Год журнала: 2025, Номер 3(1)

Опубликована: Март 13, 2025

Язык: Английский

---