Canada

Breast cancer detection relies heavily on mammographic imaging, but dense tissue can make malignant lesions harder to identify as both dense tissue and tumours appear radiographically similar. With increasing use of AI-based diagnostic tools in medical imaging, it is important to determine whether these models are affected by the same limitations that challenge human interpretation. The purpose of this research is to emphasize the need to train AI models on a more representative dataset that includes high-noise samples to ensure equitable care across demographics. This study investigates whether variability in breast tissue density acts as a confounding factor in AI-driven tumour detection thus leading to diagnostic inaccuracies. Mammographic images from the CBIS-DDSM dataset were analyzed using Mammo-CLIP, a pretrained model for mammography-based classification. Cases were grouped by BI-RADS breast density categories, and model performance was evaluated using sensitivity, specificity, accuracy, FNR, and chi-squared testing between fatty and dense breast tissue groups. Model sensitivity decreased from 64% to 34% and FNR increased from 36.1% to 66.1% with increasing breast density. Classification accuracy was significantly higher in fatty tissue than dense tissue, and cancer detection amongst malignant cases was also significantly reduced in dense tissue. These findings suggest that denser tissue negatively influences Mammo-CLIP’s ability to detect malignant lesions and increases likelihood of missed cancers. This has important clinical implications, as false negatives may delay diagnosis and treatment, particularly for patients with dense breast tissue.

Equitable Health Technologies: Overcoming Biases and False Inferences in AI Diagnostics and Drug Discovery

Effect of breast tissue density variability on AI-based tumor detection

mammography

breast cancer

artificial intelligence

detection

Welcome to the **largest national, multi-specialty research** event featuring research from nearly 700 medical students, residents, fellows, and international medical graduates.

Use the buttons below or navigation menu to access the event.
 


[![](https://assets.underline.io/uploads/markdown_image/1/image/3e33a8c754cc49610b30950617d813e7.png)](https://underline.io/events/434/sessions?searchGroup=all)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/551404d429c4e274f9a54cbdd7081918.png)](https://underline.io/events/434/posters?searchGroup=all)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/677d92f566b880cebd36905fc738fc10.png)](https://underline.io/events/434/lounge)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/c7392833c7261c3874c5b43fa2f4df19.png)](https://underline.io/events/434/help-desk)
 

#### Event Details

The 2023 American Medical Association Research Challenge includes a poster symposium hall and a semifinals hall with the top scored research. **Log in at any time throughout the event to:**

* **View posters and presentations** on a variety of topics and specialties
* **Score posters in the semifinals** to help decide the top five finalists (AMA member exclusive opportunity)
* **Connect with participants** and provide feedback to help advance their research
* **Attend education sessions** with physician experts on key research topics

At the conclusion of this event, the AMA will announce the top five finalists who will advance to the AMA Research Challenge finals, present their research to an elite panel of judges, and compete to win a $10,000 grand prize sponsored by Laurel Road. 

### Grand Prize Sponsor

AMA would like to thank Laurel Road for sponsoring a $10,000 grand prize for the winner of the AMA Research Challenge. Laurel Road is not involved in the judging or scoring process to select the winner. The winner will be announced on February 4, 2024 during the AMA Research Challenge finals. Payment of the grand prize will be directly from Laurel Road. See the AMA Research Challenge Official Rules for terms and conditions.

![](https://assets.underline.io/uploads/markdown_image/1/image/611e331d4a0eba0920f240cfc04b4d77.png)

**There is no registration fee for this event! **

Register to explore research in a variety of topics. AMA members get exclusive access to score posters in the semifinals.

**Please use this link [here](https://www.ama-assn.org/about/events/ama-research-challenge) to register for free.**

Registration Is Required

AMA Research Challenge 2023

Join the AMA Research Challenge poster symposium and semifinals to explore top research posters, score posters in the semifinals (an AMA member exclusive opportunity) and attend education sessions with experienced physicians to learn tips to advance your research and career. 

technical paper

As AI transforms healthcare, ensuring reliability is critical. However, systems learning from existing data often reflect gaps or inconsistencies, leading to false inferences, diagnostic errors, and unequal performance across patient groups.

Open Data is pivotal for transparency and accountability. Biases emerging from unrepresentative datasets or flawed assumptions can perpetuate health disparities.

Leveraging public data on genomics, clinical trials, and social determinants allows researchers to audit and mitigate these risks effectively.

We task undergraduate and graduate students to harness data science to:

* Detect, reduce, and prevent bias in diagnostics and drug discovery;

* Develop data-driven tools that increase fairness and inclusivity;

* Innovate bias-aware model evaluation techniques to build trustworthy AI systems.

As AI transforms healthcare, ensuring reliability is critical. However, systems learning from existing data often reflect gaps or inconsistencies, leading to false inferences, diagnostic errors, and unequal performance across patient groups.

Artificial Intelligence (AI) systems for skin-disease classification can perform unevenly across skin tones, risking wider disparities in dermatological care. Using Fitzpatrick17k, we trained a Swin Transformer classifier and observed a 26.4-percentage-point accuracy gap across Fitzpatrick skin-type groups. This gap was non-monotonic: accuracy was lowest for the lightest group (41.9%) and highest for a mid-tone group (68.3%), not following a simple lighter-versus-darker pattern. We then tested whether dermatology-specific transfer learning improves accuracy, fairness, and cross-dataset generalization. We standardized five public datasets (Fitzpatrick17k, DermaCon-IN, Derm1M, PASSION, DDI) to a unified eight-class taxonomy and evaluated staged transfer learning, large-scale pretraining on Derm1M, and concept-bottleneck models. Cross-dataset fine-tuning slightly improved accuracy but widened the skin-tone gap from 26.4% to 29.9%. Derm1M pretraining raised Top-5 accuracy and macro AUC by 1.6 and 7.0 percentage points but did not improve Top-1 accuracy or reduce the skin-tone gap, and external validation on PASSION and DDI showed marked performance declines, indicating limited robustness under dataset shift. Concept-bottleneck models matched baseline accuracy while adding interpretability. Persistent inflammatory-versus-infectious confusion and frequent high-confidence errors remained. These results show that taxonomy standardization and domain-specific pretraining, though necessary for combining fragmented datasets, are not sufficient to close fairness gaps; explicit fairness- and imbalance-aware methods with multi-dataset training are the needed next steps.

Skin-Tone Fairness in Dermatology AI: Limits of Pretraining and Harmonization

Healthcare systems generate vast quantities of clinical data, yet much of this information remains locked within unstructured free-text records, including physician notes and doctor-patient transcripts, making it inaccessible for large-scale analysis. Manually extracting structured information from this data is time-consuming, expensive, and error-prone. Large language models (LLMs) have demonstrated promising performance in a variety of medical fields, including radiology report summary, doctor-patient dialogue transcription, clinical question answering, and diagnostic reasoning. This study evaluates four LLMs (GPT-5.4 mini, Qwen3 32B, Llama 3.1 8B, and Gemini 3.1 Flash-Lite Preview) on their ability to label radiology diagnostic conclusions and clinical assessment summaries from unstructured medical text using MedVAL-Bench, a 2025 Stanford MIMI benchmark with physician-validated reference outputs. The models were tested on two tasks: doctor-patient dialogue summarization (dialogue2note, n=85) and radiology report impression extraction (report2impression, n=190), which were assessed using ROUGE-1, BERTScore F1, and BLEU. The results suggest that proprietary models like GPT-5.4 mini outperformed all models with a ROUGE-1 of 0.376 and BERTScore F1 of 0.860, while Qwen3 32B, Llama 3.1 8B, and Gemini 3.1 Flash-Lite Preview produced comparable results. ROUGE-1 and BLEU scores were consistently low across all models, reflecting the limitations of n-gram metrics in clinical contexts, where a single-word difference can carry critical medical significance. BERTScore F1 emerged as the most reliable indicator of semantic similarity. These findings suggest LLMs can significantly automate clinical text labelling, offering a scalable, low-cost alternative to manual data extraction in healthcare research.

A Comparative Analysis of LLM-Assisted Labelling of Unstructured Clinical Data

General-purpose artificial intelligence chatbots are increasingly used for health information, despite not being validated as medical devices. In 2026, ECRI ranked misuse of AI chatbots as the leading health technology hazard, noting that human-sounding responses can mislead users when information affects patient care. Dermatology is a particularly important domain for AI safety because patients can easily photograph visible skin concerns and seek advice before entering formal care. Although specialized dermatology models such as SkinGPT-4 demonstrate the feasibility of image-based AI analysis, widely available general-purpose tools such as ChatGPT, Claude, and Gemini may be more accessible to ordinary users. This paper proposes SAFER-Skin, a reproducible safety-audit framework for evaluating the off-label use of general-purpose multimodal chatbots when responding to consumer-style dermatology image queries. Rather than diagnostic accuracy, the framework is designed to evaluate safety behavior: whether chatbots provide appropriate referral advice, communicate uncertainty, avoid false reassurance, and refrain from prioritizing skincare products when medical evaluation may be needed. The proposed methodology uses publicly available dermatology image datasets including the Diverse Dermatology Images dataset and the Skin Condition Image Network, paired with neutral, reassurance/product-seeking, and safety-first prompt templates. Responses would be coded using a structured rubric with a primary safety classification and seven binary safety features, then stratified by model, prompt style, risk tier, dataset, and available skin-tone metadata. No empirical results are reported; this manuscript presents the proposed framework and protocol. SAFER-Skin is intended for AI safety researchers, health AI developers, and clinical governance teams. It does not validate AI for clinical diagnosis or replace dermatologists; it provides a practical method for identifying unsafe response patterns before off-label AI use in dermatology becomes entrenched.

SAFER-Skin: Auditing AI Skin Advice Across Skin Tones

Traumatic brain injury (TBI) is an important public health concern in Canada and is frequently associated with depression – although often underdiagnosed in this population. Large-scale population health datasets, like ones provided by Statistics Canada, typically rely on clinical diagnoses as outcome variables. As these datasets are increasingly used to develop artificial intelligence (AI) diagnostic models, systematic disparities in diagnosis may become embedded within algorithmic systems – contributing to existing health inequities. This study examined the relationship between TBI and diagnosed mood disorders in Canada – exploring demographic differences that may influence diagnosis and healthcare representation. Data were obtained from the Canadian Community Health Survey (CCHS) 2017–2018 Public Use Microdata File. Descriptive analyses, chi-square tests, and multivariable logistic regression compared mood disorder prevalence between respondents reporting TBI and those reporting other injuries, controlling for sex, age, and immigrant status. The final sample included 15,884 respondents, of whom 611 reported a TBI. Mood disorders were more prevalent among TBI survivors (18.6%) than among those reporting other injuries (14.3%). TBI remained significantly associated with increased odds of reporting a mood disorder (OR = 1.33, p = .008) after adjustment for demographic variables. Women with TBI reported higher rates of mood disorders than men (22.8% vs. 13.3%), and non-immigrants demonstrated greater odds of diagnosis than immigrants. These findings suggest that depression remains a significant consequence of TBI, highlighting how demographic differences in diagnosis may influence datasets used to develop AI technologies; underscoring the importance of equitable data representation in future healthcare innovation.

Bias in Depression Diagnosis Following Traumatic Brain Injury: Implications for Equitable AI Diagnostics in Canada

Clinical trials provide the evidence used to guide treatment recommendations, regulatory decisions, and healthcare practice. As clinical evidence increasingly supports precision medicine and AI-assisted healthcare, the completeness and representativeness of clinical trial data have become important considerations. This study examined demographic representation and reporting practices across cardiovascular, respiratory, and diabetes/metabolic clinical trials identified through ClinicalTrials.gov. A cross-sectional analysis of 54 completed studies was conducted, including 18 cardiovascular studies, 18 respiratory studies, and 18 diabetes/metabolic studies. Data were collected on sex distribution, race and ethnicity reporting, enrollment, and study design. Descriptive analyses compared demographic representation and reporting across disease areas. Differences were observed across all three disease categories. Cardiovascular studies contained the most missing demographic information, with 11,997 participants lacking sex-specific reporting and 11,088 participants classified as Unknown/Not Specified for race or ethnicity. Respiratory studies generally demonstrated more complete demographic reporting, while diabetes/metabolic studies reported the highest number of female participants. Interventional studies comprised 45 of the 54 included studies. These findings suggest that demographic representation and reporting practices remain inconsistent across clinical research. Incomplete demographic reporting may contribute to evidence gaps that limit the interpretation and generalizability of clinical findings. As clinical trial datasets increasingly inform precision medicine, predictive analytics, and AI-assisted healthcare systems, improving demographic inclusion and reporting standards may strengthen the quality, reliability, and applicability of future healthcare evidence and technologies.

Demographic representation and reporting in clinical trials across major disease areas

Breast cancer screening relies heavily on detailed mammographic interpretation to detect and classify lesions. Advancements in multimodal artificial intelligence (AI) systems have demonstrated potential capabilities in cancer imaging analysis. However, the accuracy and reliability of AI interpretation of mammographic lesions remain understudied. This study evaluates the diagnostic accuracy of commercial AI models for detecting and classifying mammographic lesions, compared with physician annotations. Using breast cancer patient imaging data, from the CBIS-DDSM, hosted by the TCIA. Each image case was assessed on five diagnostic criteria: lesion presence, lesion count, shape, margin, and pathology, compared against expert radiologist annotations. Across 121 individual mammograms, Gemini achieved the highest overall diagnostic accuracy (> 60%), outperforming Claude and ChatGPT. All models demonstrated high correctness for simple features ; lesion presence (≥ 90%) and lesion count (85–95%), but exhibited marked declines in complex morphological characterization, with accuracy for margin and shape ranging from 5–30%. Pathology classification achieved moderate accuracy (50–60%), reflecting difficulty in integrating interdependent visual and semantic cues. These findings reveal that while LMMs can reliably detect lesions, their interpretive precision diminishes when constrained by BI‑RADS‑defined morphological taxonomy. To address these limitations, we developed TerraComrad, a prototype human‑in‑the‑loop web application that integrates clinician‑guided region selection to improve lesion characterisation. Overall, this study establishes a benchmark for evaluating general‑purpose multimodal AI in breast imaging, demonstrating potential for preliminary screening while emphasizing the necessity of expert oversight for complex diagnostic interpretation.

Performance of commercial large multimodal models in mammographic lesion interpretation

Cardiovascular diseases remain a leading cause of morbidity, making timely and accurate diagnosis essential for improving patient outcomes. However, sex-based disparities in the diagnosis of cardiovascular disease lead to increased risk of misdiagnosis, and subsequent undertreatment, in females. As artificial intelligence (AI), machine learning (ML), and large language models (LLMs) become increasingly integrated into clinical decision-making. It is important to determine whether these tools reduce or reproduce existing biases in existing medical data. This manuscript reviews evidence on sex-based bias in AI-assisted cardiovascular diagnosis. A two-phase search strategy was used to identify relevant literature from OMNI McMaster, Google Scholarly, PubMed, and Web of Science. Initial keywords produced fifty articles, which were screened using an inclusion and exclusion criteria related to human participants, clinical cardiovascular populations, AI or ML use, and sex-stratified evaluation. Fifteen articles were selected for detailed analysis. Across the reviewed literature, findings suggest sex-based disparities exist in both traditional clinical and AI-supported diagnosis. These findings highlight the need for further sex-stratified model evaluation, more representative data, and bias-mitigation strategies before AI tools are more widely implemented. Overall, AI has the potential to improve diagnosis, but without careful training and bias control, it may exaggerate existing gender-based biases.

Biases In LLMs for Clinical Diagnosis of General Cardiovascular Disease

Accurate demographic representation in clinical trials is crucial for data generalizability, yet reporting for specific ethnicities remains under-examined. Using a Python-based algorithm, this study evaluated the visibility of South Asian participants across 178 global ribociclib HER2-negative breast cancer trial records extracted from ClinicalTrials.gov. A novel, four-level rule-based framework, the South Asian Visibility (SAV) score, was developed to quantify reporting granularity, ranging from 0 (no demographic reporting) to 3 (explicit subgroup reporting). Natural language processing (NLP) evaluated target keyword frequencies, while non-parametric tests assessed relationships between study scale, design, and visibility. The analysis revealed a severe reporting deficit: 161 studies (90.4%) provided zero public demographic data (SAV score 0), whereas only six (3.4%) explicitly isolated South Asian subgroups (SAV score 3). Furthermore, sample size demonstrated a negligible, non-significant correlation with visibility scores (r_s = 0. 072, p = 0. 3491) , confirming that larger research scales do not inherently improve data transparency. NLP analysis supported this observation, yielding zero text mentions of "Asian" or "South Asian" despite frequent references to "race" and "ethnicity." Ultimately, reliance on broad, combined racial categories severely masks South Asian representation in clinical data. This reporting gap may limit the assessment of generalizability of oncological findings and threatens the equity of downstream data-driven technologies, such as precision oncology models. Mandating subgroup-level demographic reporting standards across global clinical registries is vital to achieving equitable health outcomes.

South Asian Visibility in Ribociclib Clinical Trials for HER2-Negative Breast Cancer: An Algorithmic Assessment of Demographic Reporting

Background: Low-cost screening models may expand access to chronic disease screening, but acceptable overall performance can conceal false-negative classifications in population subgroups that may lead to false reassurance. Objective: We evaluated whether a low-cost model for the composite chronic disease outcome produced unequal false-negative rates across demographic and socioeconomic groups and acceptable overall calibration. Methods: We analyzed 5,569 adults aged 20 years or older in the 2017–2018 National Health and Nutrition Examination Survey. Predictors included demographic, socioeconomic, behavioral, self-rated health, and body mass index variables. The outcome indicated current evidence of diabetes, an elevated blood-pressure indicator, cardiovascular disease or stroke, or chronic kidney disease. Logistic regression was the primary model and random forest was the benchmark. Performance was evaluated in a stratified 30% test set. Results: The composite outcome was present in 3,601 participants (64.7%). Logistic regression achieved an AUC of 0.836 (95% CI, 0.816–0.856), sensitivity of 0.856 (0.833–0.875), specificity of 0.624 (0.585–0.662), and false-negative rate of 0.144 (0.125–0.167). Random forest achieved an AUC of 0.821 (0.800–0.842) and false-negative rate of 0.117 (0.099–0.137). Among outcome-positive adults aged 20–39 years, false-negative rates were 0.684 (0.615–0.747) for logistic regression and 0.535 (0.463–0.605) for random forest. Logistic regression had better overall calibration. Conclusion: Overall metrics concealed substantial age-related differences in false-negative rates. Screening-model evaluation should routinely report subgroup false-negative rates together with overall calibration before deployment.

Subgroup False-Negative Rates and Overall Calibration in NHANES Chronic Disease Screening

Genomic research and precision medicine increasingly rely on Genome-Wide Association Studies (GWAS) and Polygenic Risk Scores (PRS) to predict disease risk. However, the fundamental datasets that train these models are overwhelmingly composed of individuals of European descent, raising vital concerns about algorithmic bias. This study measures the extent of this systematic overrepresentation of European ancestry in global genetic databases and the risks it poses to diagnostic accuracy for non-European populations. Using a Comparative Statistical Audit and Biomarker Sensitivity Analysis, we evaluated the gene DENND1A, a major susceptibility locus for Polyendocrine Metabolic Ovarian Syndrome (PMOS), formerly known as Polycystic Ovary Syndrome (PCOS). Our analysis reveals a severe gap in the representation of different groups in genetic databases; although certain PMOS-linked genetic markers are highly prevalent in African and South Asian populations, the sequencing data collected for these groups remains vastly smaller than for European cohorts. Consequently, predictive AI algorithms trained on Eurocentric baselines can mathematically misclassify harmless, high-frequency variations in underrepresented groups as abnormal, leading to false positives. To ensure diagnostic AI does not systematically misdiagnose minority populations, achieving equitable clinical outcomes requires a fundamental shift toward ancestral diversity in foundational training data.

Downloads

Next from Equitable Health Technologies: Overcoming Biases and False Inferences in AI Diagnostics and Drug Discovery

Skin-Tone Fairness in Dermatology AI: Limits of Pretraining and Harmonization

Similar lecture

Elevated glycan expression in stage I breast cancer tumors

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from Equitable Health Technologies: Overcoming Biases and False Inferences in AI Diagnostics and Drug Discovery

Skin-Tone Fairness in Dermatology AI: Limits of Pretraining and Harmonization

Similar lecture

Elevated glycan expression in stage I breast cancer tumors

Downloads