A scientometric analysis of fairness in health AI literature.

Artificial intelligence (AI) and machine learning are central components of today's medical environment. The fairness of AI, i.e. the ability of AI to be free from bias, has repeatedly come into question. This study investigates the diversity of members of academia whose scholarship poses questions about the fairness of AI. The articles that combine the topics of fairness, artificial intelligence, and medicine were selected from Pubmed, Google Scholar, and Embase using keywords. Eligibility and data extraction from the articles were done manually and cross-checked by another author for accuracy. Articles were selected for further analysis, cleaned, and organized in Microsoft Excel; spatial diagrams were generated using Public Tableau. Additional graphs were generated using Matplotlib and Seaborn. Linear and logistic regressions were conducted using Python to measure the relationship between funding status, number of citations, and the gender demographics of the authorship team. We identified 375 eligible publications, including research and review articles concerning AI and fairness in healthcare. Analysis of the bibliographic data revealed that there is an overrepresentation of authors that are white, male, and are from high-income countries, especially in the roles of first and last author. Additionally, analysis showed that papers whose authors are based in higher-income countries were more likely to be cited more often and published in higher impact journals. These findings highlight the lack of diversity among the authors in the AI fairness community whose work gains the largest readership, potentially compromising the very impartiality that the AI fairness community is working towards.

From quality improvement to equality improvement projects: A scoping review and framework.

Increasing awareness of health disparities has led to proposals for a pay-for-equity scheme. Implementing such proposals requires systematic methods of collecting and reporting health outcomes for targeted demographics over time. This lays the foundation for a shift from quality improvement projects (QIPs) to equality improvement projects (EQIPs) that could evaluate adherence to standards and progress toward health equity. We performed a scoping review on EQIPs to inform a new framework for quality improvement through a health equity lens. Forty studies implemented an intervention after identifying a disparity compared to 149 others which merely identified group differences. Most evaluated race-based differences and were conducted at the institutional level, with representation in both the inpatient and outpatient settings. EQIPs that improved equity leveraged multidisciplinary expertise, healthcare staff education, and developed tools to track health outcomes continuously. EQIPs can help bridge the inequality gap and form part of an incentivized systematic equality improvement framework.

Familial pancreatic cancer: who should be considered for genetic testing?

BACKGROUND: Determining how many female patients who underwent breast imaging meet the eligibility criteria for genetic testing for familial pancreatic cancer (FPC). METHODS: A total of 42,904 patients seen at the Newton-Wellesley Hospital between 2007 and 2009 were retrospectively reviewed. The first four categories were based on pancreatic cancer-associated syndromes: (1) hereditary breast and ovarian cancer (HBOC), (2) Lynch syndrome (LS), (3) familial atypical multiple mole melanoma (FAMMM), and (4) family history of FPC (FH-FPC). PancPRO (5) and MelaPRO (6) categories were based on risk scores from Mendelian risk prediction tool. RESULTS: Exactly 4445 of 42,904 patients were found to be in at least one of the six risk categories. About 5.7% of patients were classified as being at high risk for HBOC, 2.3% as being at high risk for LS, 0.1% as being at high risk for FAMMM, 0.1% as being at high risk for FH-FPC, 2.7% as being at high risk based on PancPRO, and 0.2% as being at high risk based on MelaPRO. CONCLUSION: About 10.4% of the female patients were classified as being at high risk for FPC. This finding emphasizes the importance of applying criteria to the general population, in order to ensure that individuals with high risk are identified early.

The Incidence of Node-Positive Non-small-Cell Lung Cancer Undergoing Sublobar Resection and the Role of Radiation in Its Management.

Purpose: To identify the incidence, preoperative risk factors, and prognosis associated with pathologically positive lymph node (pN+) in patients undergoing a sub-lobar resection (SLR). Methods: This is a retrospective study using the National Cancer Database (NCDB) from 2004 to 2014 analyzing SLR excluding those with any preoperative chemotherapy and/or radiation, follow-up <3 months, stage IV disease, or >1 tumor nodule. Multivariable modeling (MVA) was used to determine factors associated with overall survival (OS). Propensity score matching (PSM) was used to determine preoperative risk factors for pN+ in patients having at least one node examined to assess radiation's effect on OS in those patients with pN+ and to determine whether SLR was associated with inferior OS as compared to lobectomy for each nodal stage. Results: A total of 40,202 patients underwent SLR, but only 58.3% had one lymph node examined. Then, 2,615 individuals had pN+ which decreased progressively from 15.1% in 2004 to 8.9% in 2014 (N1, from 6.3 to 3.0%, and N2, from 8.4 to 5.9%). A lower risk of pN+ was noted for squamous cell carcinomas, bronchioloalveolar adenocarcinoma (BAC), adenocarcinomas, and right upper lobe locations. In the pN+ group, OS was worse without chemotherapy or radiation. Radiation was associated with a strong trend for OS in the entire pN+ group (p = 0.0647) which was largely due to the effects on those having N2 disease (p = 0.009) or R1 resections (p = 0.03), but not N1 involvement (p = 0.87). PSM noted that SLR was associated with an inferior OS as compared to lobectomy by nodal stage in the overall patient population and even for those with tumors <2 cm. Conclusion: pN+ incidence in SLRs has decreased over time. SLR was associated with inferior OS as compared to lobectomy by nodal stage. Radiation appears to improve the OS in patients undergoing SLR with pN+, especially in those with N2 nodal involvement and/or positive margins.

The role of Micro-CT in imaging breast cancer specimens.

PURPOSE: The goal of breast cancer surgery is to remove all of the cancer with a minimum of normal tissue, but absence of full 3-dimensional information on the specimen makes this difficult to achieve. METHOD: Micro-CT is a high resolution, X-ray, 3D imaging method, widely used in industry but rarely in medicine. RESULTS: We imaged and analyzed 173 partial mastectomies (129 ductal carcinomas, 14 lobular carcinomas, 28 DCIS). Imaging was simple and rapid. The size and shape of the cancers seen on Micro-CT closely matched the size and shape of the cancers seen at specimen dissection. Micro-CT images of multicentric/multifocal cancers revealed multiple non-contiguous masses. Micro-CT revealed cancer touching the specimen edge for 93% of the 114 cases judged margin positive by the pathologist, and 28 of the cases not seen as margin positive on pathological analysis; cancer occupied 1.55% of surface area when both the pathologist and Micro-CT suggested cancer at the edge, but only 0.45% of surface area for the "Micro-CT-Only-Positive Cases". Thus, Micro-CT detects cancers that touch a very small region of the specimen surface, which is likely to be missed on sectioning. CONCLUSIONS: Micro-CT provides full 3D images of breast cancer specimens, allowing one to identify, in minutes rather than hours, while the patient is in OR, margin-positive cancers together with information on where the cancer touches the edge, in a fashion more accurate than possible from the histology slides alone.

Performance of Breast Cancer Risk-Assessment Models in a Large Mammography Cohort.

BACKGROUND: Several breast cancer risk-assessment models exist. Few studies have evaluated predictive accuracy of multiple models in large screening populations. METHODS: We evaluated the performance of the BRCAPRO, Gail, Claus, Breast Cancer Surveillance Consortium (BCSC), and Tyrer-Cuzick models in predicting risk of breast cancer over 6 years among 35 921 women aged 40-84 years who underwent mammography screening at Newton-Wellesley Hospital from 2007 to 2009. We assessed model discrimination using the area under the receiver operating characteristic curve (AUC) and assessed calibration by comparing the ratio of observed-to-expected (O/E) cases. We calculated the square root of the Brier score and positive and negative predictive values of each model. RESULTS: Our results confirmed the good calibration and comparable moderate discrimination of the BRCAPRO, Gail, Tyrer-Cuzick, and BCSC models. The Gail model had slightly better O/E ratio and AUC (O/E = 0.98, 95% confidence interval [CI] = 0.91 to 1.06, AUC = 0.64, 95% CI = 0.61 to 0.65) compared with BRCAPRO (O/E = 0.94, 95% CI = 0.88 to 1.02, AUC = 0.61, 95% CI = 0.59 to 0.63) and Tyrer-Cuzick (version 8, O/E = 0.84, 95% CI = 0.79 to 0.91, AUC = 0.62, 95% 0.60 to 0.64) in the full study population, and the BCSC model had the highest AUC among women with available breast density information (O/E = 0.97, 95% CI = 0.89 to 1.05, AUC = 0.64, 95% CI = 0.62 to 0.66). All models had poorer predictive accuracy for human epidermal growth factor receptor 2 positive and triple-negative breast cancers than hormone receptor positive human epidermal growth factor receptor 2 negative breast cancers. CONCLUSIONS: In a large cohort of patients undergoing mammography screening, existing risk prediction models had similar, moderate predictive accuracy and good calibration overall. Models that incorporate additional genetic and nongenetic risk factors and estimate risk of tumor subtypes may further improve breast cancer risk prediction.

A Clinical Decision Support Tool to Predict Cancer Risk for Commonly Tested Cancer-Related Germline Mutations.

The rapid drop in the cost of DNA sequencing led to the availability of multi-gene panels, which test 25 or more cancer susceptibility genes for a low cost. Clinicians and genetic counselors need a tool to interpret results, understand risk of various cancers, and advise on a management strategy. This is challenging as there are multiple studies regarding each gene, and it is not possible for clinicians and genetic counselors to be aware of all publications, nor to appreciate the relative accuracy and importance of each. Through an extensive literature review, we have identified reliable studies and derived estimates of absolute risk. We have also developed a systematic mechanism and informatics tools for (1) data curation, (2) the evaluation of quality of studies, and (3) the statistical analysis necessary to obtain risk. We produced the risk prediction clinical decision support tool ASK2ME (All Syndromes Known to Man Evaluator). It provides absolute cancer risk predictions for various hereditary cancer susceptibility genes. These predictions are specific to patients' gene carrier status, age, and history of relevant prophylactic surgery. By allowing clinicians to enter patient information and receive patient-specific cancer risks, this tool aims to have a significant impact on the quality of precision cancer prevention and disease management activities relying on panel testing. It is important to note that this tool is dynamic and constantly being updated, and currently, some of its limitations include (1) for many gene-cancer associations risk estimates are based on one study rather than meta-analysis, (2) strong assumptions on prior cancers, (3) lack of uncertainty measures, and (4) risk estimates for a growing set of gene-cancer associations which are not always variant specific. All of these concerns are being addressed on an ongoing basis, aiming to make the tool even more accurate.

Machine learning to parse breast pathology reports in Chinese.

INTRODUCTION: Large structured databases of pathology findings are valuable in deriving new clinical insights. However, they are labor intensive to create and generally require manual annotation. There has been some work in the bioinformatics community to support automating this work via machine learning in English. Our contribution is to provide an automated approach to construct such structured databases in Chinese, and to set the stage for extraction from other languages. METHODS: We collected 2104 de-identified Chinese benign and malignant breast pathology reports from Hunan Cancer Hospital. Physicians with native Chinese proficiency reviewed the reports and annotated a variety of binary and numerical pathologic entities. After excluding 78 cases with a bilateral lesion in the same report, 1216 cases were used as a training set for the algorithm, which was then refined by 405 development cases. The Natural language processing algorithm was tested by using the remaining 405 cases to evaluate the machine learning outcome. The model was used to extract 13 binary entities and 8 numerical entities. RESULTS: When compared to physicians with native Chinese proficiency, the model showed a per-entity accuracy from 91 to 100% for all common diagnoses on the test set. The overall accuracy of binary entities was 98% and of numerical entities was 95%. In a per-report evaluation for binary entities with more than 100 training cases, 85% of all the testing reports were completely correct and 11% had an error in 1 out of 22 entities. CONCLUSION: We have demonstrated that Chinese breast pathology reports can be automatically parsed into structured data using standard machine learning approaches. The results of our study demonstrate that techniques effective in parsing English reports can be scaled to other languages.

The impact of patient age on breast cancer risk prediction models.

BACKGROUND: The impact of age on breast cancer risk model calculations at the population level has not been well documented. METHODS: Retrospective analysis of formal breast cancer risk assessment in 36 542 females ages 40-84 at a single institution from 02/2007 to 12/2009. Five-year and lifetime breast cancer risks were calculated using Gail, Tyrer-Cuzick version 6 (TC6), Tyrer-Cuzick version 7 (TC7), BRCAPRO, and Claus models. Risk of BRCA mutation was calculated using BRCAPRO, TC6, TC7, and Myriad. Eligibility for BRCA testing was assessed using NCCN guidelines. Descriptive analyses were performed and trends in risk were assessed by age. RESULTS: The lifetime risk of breast cancer trended down with increasing age in all risk models. TC7 calculated the highest estimates for lifetime risk for all age ranges and had the highest proportion of patients with a calculated lifetime risk >20%. Five-year risk increased with age in all models. By age 60-64, every risk model predicted a mean 5-year risk ≥1.7%. Myriad estimated >5% risk of BRCA mutation more often than other models for all ages. Risk of BRCA mutation stayed constant with age with Myriad, but trended down with increasing age with TC6, TC7, and BRCAPRO. CONCLUSIONS: More patients have an estimated lifetime risk of breast cancer >20% and qualify for MRI screening with the Tyrer-Cuzick model. All models predict an increased 5-year risk with age, which could impact chemoprevention recommendations. To maximize access to genetic testing, the Myriad model and NCCN guidelines should be used.

Reassessing risk models for atypical hyperplasia: age may not matter.

PURPOSE: The aim of this study was to investigate the influence of age at diagnosis of atypical hyperplasia ("atypia", ductal [ADH], lobular [ALH], or severe ADH) on the risk of developing subsequent invasive breast cancer or ductal carcinoma in situ (DCIS). METHODS: Using standard survival analysis methods, we retrospectively analyzed 1353 women not treated with chemoprevention among a cohort of 2370 women diagnosed with atypical hyperplasia to determine the risk relationship between age at diagnosis and subsequent breast cancer. RESULTS: For all atypia diagnoses combined, our cohort showed a 5-, 10-, and 15-year risk of invasive breast cancer or DCIS of 0.56, 1.25, and 1.30, respectively, with no significant difference in the (65,75] year age group. For women aged (35,75] years, we observed no significant difference in the 15-year risk of invasive breast cancer or DCIS after atypical hyperplasia, although the baseline risk for a 40-year-old woman is approximately 1/8 the risk of a 70-year-old woman. The risks associated with invasive breast cancer or DCIS for women in our cohort diagnosed with ADH, severe ADH, or ALH, regardless of age, were 7.6% (95% CI 5.9-9.3%) at 5 years, 25.1% (20.7-29.2%) at 10 years, and 40.1% (32.8-46.6%) at 15 years. CONCLUSION: In contrast to current risk prediction models (e.g., Gail, Tyrer-Cuzick) which assume that the risk of developing breast cancer increases in relation to age at diagnosis of atypia, we found the 15-year cancer risk in our cohort was not significantly different for women between the ages of 35 (excluded) and 75. This implies that the "hits" received by the breast tissue along the "high-risk pathway" to cancer might possibly supersede other factors such as age.

Breast cancer risk models: a comprehensive overview of existing models, validation, and clinical applications.

Numerous models have been developed to quantify the combined effect of various risk factors to predict either risk of developing breast cancer, risk of carrying a high-risk germline genetic mutation, specifically in the BRCA1 and BRCA2 genes, or the risk of both. These breast cancer risk models can be separated into those that utilize mainly hormonal and environmental factors and those that focus more on hereditary risk. Given the wide range of models from which to choose, understanding what each model predicts, the populations for which each is best suited to provide risk estimations, the current validation and comparative studies that have been performed for each model, and how to apply them practically is important for clinicians and researchers seeking to utilize risk models in their practice. This review provides a comprehensive guide for those seeking to understand and apply breast cancer risk models by summarizing the majority of existing breast cancer risk prediction models including the risk factors they incorporate, the basic methodology in their development, the information each provides, their strengths and limitations, relevant validation studies, and how to access each for clinical or investigative purposes.

Using machine learning to parse breast pathology reports.

PURPOSE: Extracting information from electronic medical record is a time-consuming and expensive process when done manually. Rule-based and machine learning techniques are two approaches to solving this problem. In this study, we trained a machine learning model on pathology reports to extract pertinent tumor characteristics, which enabled us to create a large database of attribute searchable pathology reports. This database can be used to identify cohorts of patients with characteristics of interest. METHODS: We collected a total of 91,505 breast pathology reports from three Partners hospitals: Massachusetts General Hospital, Brigham and Women's Hospital, and Newton-Wellesley Hospital, covering the period from 1978 to 2016. We trained our system with annotations from two datasets, consisting of 6295 and 10,841 manually annotated reports. The system extracts 20 separate categories of information, including atypia types and various tumor characteristics such as receptors. We also report a learning curve analysis to show how much annotation our model needs to perform reasonably. RESULTS: The model accuracy was tested on 500 reports that did not overlap with the training set. The model achieved accuracy of 90% for correctly parsing all carcinoma and atypia categories for a given patient. The average accuracy for individual categories was 97%. Using this classifier, we created a database of 91,505 parsed pathology reports. CONCLUSIONS: Our learning curve analysis shows that the model can achieve reasonable results even when trained on a few annotations. We developed a user-friendly interface to the database that allows physicians to easily identify patients with target characteristics and export the matching cohort. This model has the potential to reduce the effort required for analyzing large amounts of data from medical records, and to minimize the cost and time required to glean scientific insight from these data.

What's New in Genetic Testing for Cancer Susceptibility?

The advent of next-generation sequencing, and its transition further into the clinic with the US Food and Drug Administration approval of a cystic fibrosis assay in 2013, have increased the speed and reduced the cost of DNA sequencing. Coupled with a historic ruling by the Supreme Court of the United States that human genes are not patentable, these events have caused a seismic shift in genetic testing in clinical medicine. More labs are offering genetic testing services; more multigene panels are available for gene testing; more genes and gene mutations are being identified; and more variants of uncertain significance, which may or may not be clinically actionable, have been found. All these factors, taken together, are increasing the complexity of clinical management. While these developments have led to a greater interest in genetic testing, risk assessment, and large-scale population screening, they also present unique challenges. The dilemma for clinicians is how best to understand and manage this rapidly growing body of information to improve patient care. With millions of genetic variants of potential clinical significance and thousands of genes associated with rare but well-established genetic conditions, the complexities of genetic data management clearly will require improved computerized clinical decision support tools, as opposed to continued reliance on traditional rote, memory-based medicine.

The Role of Micro-CT in 3D Histology Imaging.

OBJECTIVES: 3D histology tissue modeling is a useful analytical technique for understanding anatomy and disease at the cellular level. However, the current accuracy of 3D histology technology is largely unknown, and errors, misalignment and missing information are common in 3D tissue reconstruction. We used micro-CT imaging technology to better understand these issues and the relationship between fresh tissue and its 3D histology counterpart. METHODS: We imaged formalin-fixed and 2% Lugol-stained mouse brain, human uterus and human lung tissue with micro-CT. We then conducted image analyses on the tissues before and after paraffin embedding using 3D Slicer and ImageJ software to understand how tissue changes between the fixation and embedding steps. RESULTS: We found that all tissue samples decreased in volume by 19.2-61.5% after embedding, that micro-CT imaging can be used to assess the integrity of tissue blocks, and that micro-CT analysis can help to design an optimized tissue-sectioning protocol. CONCLUSIONS: Micro-CT reference data help to identify where and to what extent tissue was lost or damaged during slide production, provides valuable anatomical information for reconstructing missing parts of a 3D tissue model, and aids in correcting reconstruction errors when fitting the image information in vivo and ex vivo.

Prediction of primary breast cancer size and T-stage using micro-computed tomography in lumpectomy specimens.

BACKGROUND: Histopathology is the only accepted method to measure and stage the breast tumor size. However, there is a need to find another method to measure and stage the tumor size when the pathological assessment is not available. Micro-computed tomography. (micro-CT) has the ability to measure tumor in three dimensions in an intact lumpectomy specimen. In this study, we aimed to determine the accuracy of micro-CT to measure and stage the primary tumor size in breast lumpectomy specimens, as compared to the histopathology. MATERIALS AND METHODS: Seventy-two women who underwent lumpectomy surgery at the Massachusetts General Hospital Department of Surgery from June 2011 to September 2011, and from August 2013 to December 2013 participated in this study. The lumpectomy specimens were scanned using micro-CT followed by routine pathological processing. The maximum dimension of the invasive breast tumor was obtained from the micro-CT image and was compared to the corresponding pathology report for each subject. RESULTS: The invasive tumor size measurement by micro-CT was underestimated in 24 cases. (33%), overestimated in 37 cases. (51%), and matched it exactly in 11 cases. (15%) compared to the histopathology measurement for all the cases. However, micro-CT T-stage classification differed from histopathology in only 11. (15.2%) with 6 cases. (8.3%) classified as a higher stage by micro-CT, and 5 cases. (6.9%) classified as lower compared to histopathology. In addition, micro-CT demonstrated a statically significant strong agreement (κ =0.6, P < 0.05) with pathological tumor size and staging for invasive ductal carcinoma. (IDC) group. In contrast, there was no agreement. (κ = -2, P = 0.67) between micro-CT and pathology in estimating and staging tumor size for invasive lobular carcinoma. (ILC) group. This could be explained by a small sample size. (7) for ILC group. CONCLUSIONS: Micro-CT is a promising modality for measuring and staging the IDC.