Joseph Y. Lo

Professor

Duke University
Country flagUnited States

Research Interests

ApplyKite TwinLearns as you apply

Make your next move.

Prepare your application

Review drafts before you send them.

Explore related searches

Articles (3)

Progression from pre-cancers like ductal carcinoma in situ (DCIS) to invasive disease (cancer) is driven by somatic evolution and is altered by clinical interventions. We hypothesized that genetic and/or phenotypic intra-tumor heterogeneity would predict clinical outcomes for DCIS since it serves as the substrate for natural selection among cells. We profiled two samples from two geographically distinct foci from each DCIS in both cross-sectional (N = 119) and longitudinal cohorts (N = 224), with whole exome sequencing, low-pass whole genome sequencing, and a panel of immunohistochemical markers. In the longitudinal cohorts, the only statistically significant predictors of time to non-invasive DCIS recurrence were the combination of treatment (lumpectomy only vs mastectomy or lumpectomy with radiation, HR = 12.13, p = 0.003, Wald test with FDR correction), ER status (HR = 0.16 for ER+ compared to ER-, p = 0.0045), and divergence in SNVs between the two samples (HR = 1.33 per 10% divergence, p = 0.018). SNV divergence also distinguished between pure DCIS and DCIS synchronous with invasive disease in the cross-sectional cohort. In contrast, the only statistically significant predictors of time to progression to invasive disease were the combination of the width of the surgical margin (HR = 0.67 per mm, p = 0.043) and the number of mutations that were detectable at high allele frequencies (HR = 1.30 per 10 SNVs, p = 0.02). These results imply that recurrence with DCIS is a clinical and biological process different from invasive progression. Significance Evolutionary measures of breast pre-cancers associate with local recurrence after surgery, as well as progression to cancer. Recurrence and progression are different biological processes impacted differently by clinical interventions.

Year:

2024

Objectives To assess the performance bias caused by sampling data into training and test sets in a mammography radiomics study. Methods Mammograms from 700 women were used to study upstaging of ductal carcinoma in situ. The dataset was repeatedly shuffled and split into training (n = 400) and test cases (n = 300) forty times. For each split, cross-validation was used for training, followed by an assessment of the test set. Logistic regression with regularization and support vector machine were used as the machine learning classifiers. For each split and classifier type, multiple models were created based on radiomics and/or clinical features. Results Area under the curve (AUC) performances varied considerably across the different data splits (e.g., radiomics regression model: train 0.58–0.70, test 0.59–0.73). Performances for regression models showed a tradeoff where better training led to worse testing and vice versa. Cross-validation over all cases reduced this variability, but required samples of 500+ cases to yield representative estimates of performance. Conclusions In medical imaging, clinical datasets are often limited to relatively small size. Models built from different training sets may not be representative of the whole dataset. Depending on the selected data split and model, performance bias could lead to inappropriate conclusions that might influence the clinical significance of the findings. Advances in knowledge Performance bias can result from model testing when using limited datasets. Optimal strategies for test set selection should be developed to ensure study conclusions are appropriate.

Year:

2024

Year:

2023

Collaborators (4)

Carlo Maley

Professor

Arizona State University

UNITED STATES
View details

E. Shelley Hwang

Duke University

UNITED STATES
View details

Aziz Khan

Assistant Professor

Mohamed Bin Zayed University of Artificial Intelligence

UNITED ARAB EMIRATES
View details

Christina Curtis

Professor of Medicine & Genetics

Stanford University

UNITED STATES
View details