AI and machine learning in medical imaging: key points from development to translation.

Innovation in medical imaging artificial intelligence (AI)/machine learning (ML) demands extensive data collection, algorithmic advancements, and rigorous performance assessments encompassing aspects such as generalizability, uncertainty, bias, fairness, trustworthiness, and interpretability. Achieving widespread integration of AI/ML algorithms into diverse clinical tasks will demand a steadfast commitment to overcoming issues in model design, development, and performance assessment. The complexities of AI/ML clinical translation present substantial challenges, requiring engagement with relevant stakeholders, assessment of cost-effectiveness for user and patient benefit, timely dissemination of information relevant to robust functioning throughout the AI/ML lifecycle, consideration of regulatory compliance, and feedback loops for real-world performance evidence. This commentary addresses several hurdles for the development and adoption of AI/ML technologies in medical imaging. Comprehensive attention to these underlying and often subtle factors is critical not only for tackling the challenges but also for exploring novel opportunities for the advancement of AI in radiology.

Accelerating Genome- and Phenome-Wide Association Studies using GPUs - A case study using data from the Million Veteran Program.

The expansion of biobanks has significantly propelled genomic discoveries yet the sheer scale of data within these repositories poses formidable computational hurdles, particularly in handling extensive matrix operations required by prevailing statistical frameworks. In this work, we introduce computational optimizations to the SAIGE (Scalable and Accurate Implementation of Generalized Mixed Model) algorithm, notably employing a GPU-based distributed computing approach to tackle these challenges. We applied these optimizations to conduct a large-scale genome-wide association study (GWAS) across 2,068 phenotypes derived from electronic health records of 635,969 diverse participants from the Veterans Affairs (VA) Million Veteran Program (MVP). Our strategies enabled scaling up the analysis to over 6,000 nodes on the Department of Energy (DOE) Oak Ridge Leadership Computing Facility (OLCF) Summit High-Performance Computer (HPC), resulting in a 20-fold acceleration compared to the baseline model. We also provide a Docker container with our optimizations that was successfully used on multiple cloud infrastructures on UK Biobank and All of Us datasets where we showed significant time and cost benefits over the baseline SAIGE model.

Artificial intelligence in medicine: mitigating risks and maximizing benefits via quality assurance, quality control, and acceptance testing.

The adoption of artificial intelligence (AI) tools in medicine poses challenges to existing clinical workflows. This commentary discusses the necessity of context-specific quality assurance (QA), emphasizing the need for robust QA measures with quality control (QC) procedures that encompass (1) acceptance testing (AT) before clinical use, (2) continuous QC monitoring, and (3) adequate user training. The discussion also covers essential components of AT and QA, illustrated with real-world examples. We also highlight what we see as the shared responsibility of manufacturers or vendors, regulators, healthcare systems, medical physicists, and clinicians to enact appropriate testing and oversight to ensure a safe and equitable transformation of medicine through AI.

Deep learning uncertainty quantification for clinical text classification.

INTRODUCTION: Machine learning algorithms are expected to work side-by-side with humans in decision-making pipelines. Thus, the ability of classifiers to make reliable decisions is of paramount importance. Deep neural networks (DNNs) represent the state-of-the-art models to address real-world classification. Although the strength of activation in DNNs is often correlated with the network's confidence, in-depth analyses are needed to establish whether they are well calibrated. METHOD: In this paper, we demonstrate the use of DNN-based classification tools to benefit cancer registries by automating information extraction of disease at diagnosis and at surgery from electronic text pathology reports from the US National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) population-based cancer registries. In particular, we introduce multiple methods for selective classification to achieve a target level of accuracy on multiple classification tasks while minimizing the rejection amount-that is, the number of electronic pathology reports for which the model's predictions are unreliable. We evaluate the proposed methods by comparing our approach with the current in-house deep learning-based abstaining classifier. RESULTS: Overall, all the proposed selective classification methods effectively allow for achieving the targeted level of accuracy or higher in a trade-off analysis aimed to minimize the rejection rate. On in-distribution validation and holdout test data, with all the proposed methods, we achieve on all tasks the required target level of accuracy with a lower rejection rate than the deep abstaining classifier (DAC). Interpreting the results for the out-of-distribution test data is more complex; nevertheless, in this case as well, the rejection rate from the best among the proposed methods achieving 97% accuracy or higher is lower than the rejection rate based on the DAC. CONCLUSIONS: We show that although both approaches can flag those samples that should be manually reviewed and labeled by human annotators, the newly proposed methods retain a larger fraction and do so without retraining-thus offering a reduced computational cost compared with the in-house deep learning-based abstaining classifier.

The United States Department of Energy and National Institutes of Health Collaboration: Medical Care Advances via Discovery in Physical Sciences.

Over several months, representatives from the U.S. Department of Energy (DOE) Office of Science and National Institutes of Health (NIH) had a number of meetings that lead to the conclusion that innovations in the Nation's health care could be realized by more directed interactions between NIH and DOE. It became clear that the expertise amassed and instrumentation advances developed at the DOE physical science laboratories to enable cutting-edge research in particle physics could also feed innovation in medical healthcare. To meet their scientific mission, the DOE laboratories created advances in such technologies as particle beam generation, radioisotope production, high-energy particle detection and imaging, superconducting particle accelerators, superconducting magnets, cryogenics, high-speed electronics, artificial intelligence, and big data. To move forward, NIH and DOE initiated the process of convening a joint workshop which occurred on July 12th and 13th, 2021. This Special Report presents a summary of the findings of the collaborative workshop and introduces the goals of the next one.

AAPM task group report 273: Recommendations on best practices for AI and machine learning for computer-aided diagnosis in medical imaging.

Rapid advances in artificial intelligence (AI) and machine learning, and specifically in deep learning (DL) techniques, have enabled broad application of these methods in health care. The promise of the DL approach has spurred further interest in computer-aided diagnosis (CAD) development and applications using both "traditional" machine learning methods and newer DL-based methods. We use the term CAD-AI to refer to this expanded clinical decision support environment that uses traditional and DL-based AI methods. Numerous studies have been published to date on the development of machine learning tools for computer-aided, or AI-assisted, clinical tasks. However, most of these machine learning models are not ready for clinical deployment. It is of paramount importance to ensure that a clinical decision support tool undergoes proper training and rigorous validation of its generalizability and robustness before adoption for patient care in the clinic. To address these important issues, the American Association of Physicists in Medicine (AAPM) Computer-Aided Image Analysis Subcommittee (CADSC) is charged, in part, to develop recommendations on practices and standards for the development and performance assessment of computer-aided decision support systems. The committee has previously published two opinion papers on the evaluation of CAD systems and issues associated with user training and quality assurance of these systems in the clinic. With machine learning techniques continuing to evolve and CAD applications expanding to new stages of the patient care process, the current task group report considers the broader issues common to the development of most, if not all, CAD-AI applications and their translation from the bench to the clinic. The goal is to bring attention to the proper training and validation of machine learning algorithms that may improve their generalizability and reliability and accelerate the adoption of CAD-AI systems for clinical decision support.

Using ensembles and distillation to optimize the deployment of deep learning models for the classification of electronic cancer pathology reports.

Objective: We aim to reduce overfitting and model overconfidence by distilling the knowledge of an ensemble of deep learning models into a single model for the classification of cancer pathology reports. Materials and Methods: We consider the text classification problem that involves 5 individual tasks. The baseline model consists of a multitask convolutional neural network (MtCNN), and the implemented ensemble (teacher) consists of 1000 MtCNNs. We performed knowledge transfer by training a single model (student) with soft labels derived through the aggregation of ensemble predictions. We evaluate performance based on accuracy and abstention rates by using softmax thresholding. Results: The student model outperforms the baseline MtCNN in terms of abstention rates and accuracy, thereby allowing the model to be used with a larger volume of documents when deployed. The highest boost was observed for subsite and histology, for which the student model classified an additional 1.81% reports for subsite and 3.33% reports for histology. Discussion: Ensemble predictions provide a useful strategy for quantifying the uncertainty inherent in labeled data and thereby enable the construction of soft labels with estimated probabilities for multiple classes for a given document. Training models with the derived soft labels reduce model confidence in difficult-to-classify documents, thereby leading to a reduction in the number of highly confident wrong predictions. Conclusions: Ensemble model distillation is a simple tool to reduce model overconfidence in problems with extreme class imbalance and noisy datasets. These methods can facilitate the deployment of deep learning models in high-risk domains with low computational resources where minimizing inference time is required.

Optimal vocabulary selection approaches for privacy-preserving deep NLP model training for information extraction and cancer epidemiology.

BACKGROUND: With the use of artificial intelligence and machine learning techniques for biomedical informatics, security and privacy concerns over the data and subject identities have also become an important issue and essential research topic. Without intentional safeguards, machine learning models may find patterns and features to improve task performance that are associated with private personal information. OBJECTIVE: The privacy vulnerability of deep learning models for information extraction from medical textural contents needs to be quantified since the models are exposed to private health information and personally identifiable information. The objective of the study is to quantify the privacy vulnerability of the deep learning models for natural language processing and explore a proper way of securing patients' information to mitigate confidentiality breaches. METHODS: The target model is the multitask convolutional neural network for information extraction from cancer pathology reports, where the data for training the model are from multiple state population-based cancer registries. This study proposes the following schemes to collect vocabularies from the cancer pathology reports; (a) words appearing in multiple registries, and (b) words that have higher mutual information. We performed membership inference attacks on the models in high-performance computing environments. RESULTS: The comparison outcomes suggest that the proposed vocabulary selection methods resulted in lower privacy vulnerability while maintaining the same level of clinical task performance.

A Keyword-Enhanced Approach to Handle Class Imbalance in Clinical Text Classification.

Recent applications ofdeep learning have shown promising results for classifying unstructured text in the healthcare domain. However, the reliability of models in production settings has been hindered by imbalanced data sets in which a small subset of the classes dominate. In the absence of adequate training data, rare classes necessitate additional model constraints for robust performance. Here, we present a strategy for incorporating short sequences of text (i.e. keywords) into training to boost model accuracy on rare classes. In our approach, we assemble a set of keywords, including short phrases, associated with each class. The keywords are then used as additional data during each batch of model training, resulting in a training loss that has contributions from both raw data and keywords. We evaluate our approach on classification of cancer pathology reports, which shows a substantial increase in model performance for rare classes. Furthermore, we analyze the impact of keywords on model output probabilities for bigrams, providing a straightforward method to identify model difficulties for limited training data.

Predictive Radiation Oncology - A New NCI-DOE Scientific Space and Community.

With a widely attended virtual kickoff event on January 29, 2021, the National Cancer Institute (NCI) and the Department of Energy (DOE) launched a series of 4 interactive, interdisciplinary workshops-and a final concluding "World Café" on March 29, 2021-focused on advancing computational approaches for predictive oncology in the clinical and research domains of radiation oncology. These events reflect 3,870 human hours of virtual engagement with representation from 8 DOE national laboratories and the Frederick National Laboratory for Cancer Research (FNL), 4 research institutes, 5 cancer centers, 17 medical schools and teaching hospitals, 5 companies, 5 federal agencies, 3 research centers, and 27 universities. Here we summarize the workshops by first describing the background for the workshops. Participants identified twelve key questions-and collaborative parallel ideas-as the focus of work going forward to advance the field. These were then used to define short-term and longer-term "Blue Sky" goals. In addition, the group determined key success factors for predictive oncology in the context of radiation oncology, if not the future of all of medicine. These are: cross-discipline collaboration, targeted talent development, development of mechanistic mathematical and computational models and tools, and access to high-quality multiscale data that bridges mechanisms to phenotype. The workshop participants reported feeling energized and highly motivated to pursue next steps together to address the unmet needs in radiation oncology specifically and in cancer research generally and that NCI and DOE project goals align at the convergence of radiation therapy and advanced computing.

Class imbalance in out-of-distribution datasets: Improving the robustness of the TextCNN for the classification of rare cancer types.

In the last decade, the widespread adoption of electronic health record documentation has created huge opportunities for information mining. Natural language processing (NLP) techniques using machine and deep learning are becoming increasingly widespread for information extraction tasks from unstructured clinical notes. Disparities in performance when deploying machine learning models in the real world have recently received considerable attention. In the clinical NLP domain, the robustness of convolutional neural networks (CNNs) for classifying cancer pathology reports under natural distribution shifts remains understudied. In this research, we aim to quantify and improve the performance of the CNN for text classification on out-of-distribution (OOD) datasets resulting from the natural evolution of clinical text in pathology reports. We identified class imbalance due to different prevalence of cancer types as one of the sources of performance drop and analyzed the impact of previous methods for addressing class imbalance when deploying models in real-world domains. Our results show that our novel class-specialized ensemble technique outperforms other methods for the classification of rare cancer types in terms of macro F1 scores. We also found that traditional ensemble methods perform better in top classes, leading to higher micro F1 scores. Based on our findings, we formulate a series of recommendations for other ML practitioners on how to build robust models with extremely imbalanced datasets in biomedical NLP applications.

Artificial intelligence in cancer research, diagnosis and therapy.

STANDFIRST: Artificial intelligence and machine learning techniques are breaking into biomedical research and health care, which importantly includes cancer research and oncology, where the potential applications are vast. These include detection and diagnosis of cancer, subtype classification, optimization of cancer treatment and identification of new therapeutic targets in drug discovery. While big data used to train machine learning models may already exist, leveraging this opportunity to realize the full promise of artificial intelligence in both the cancer research space and the clinical space will first require significant obstacles to be surmounted. In this Viewpoint article, we asked four experts for their opinions on how we can begin to implement artificial intelligence while ensuring standards are maintained so as transform cancer diagnosis and the prognosis and treatment of patients with cancer and to drive biological discovery.

Pharmacoepidemiology, Machine Learning, and COVID-19: An Intent-to-Treat Analysis of Hydroxychloroquine, With or Without Azithromycin, and COVID-19 Outcomes Among Hospitalized US Veterans.

Hydroxychloroquine (HCQ) was proposed as an early therapy for coronavirus disease 2019 (COVID-19) after in vitro studies indicated possible benefit. Previous in vivo observational studies have presented conflicting results, though recent randomized clinical trials have reported no benefit from HCQ among patients hospitalized with COVID-19. We examined the effects of HCQ alone and in combination with azithromycin in a hospitalized population of US veterans with COVID-19, using a propensity score-adjusted survival analysis with imputation of missing data. According to electronic health record data from the US Department of Veterans Affairs health care system, 64,055 US Veterans were tested for the virus that causes COVID-19 between March 1, 2020 and April 30, 2020. Of the 7,193 veterans who tested positive, 2,809 were hospitalized, and 657 individuals were prescribed HCQ within the first 48-hours of hospitalization for the treatment of COVID-19. There was no apparent benefit associated with HCQ receipt, alone or in combination with azithromycin, and there was an increased risk of intubation when HCQ was used in combination with azithromycin (hazard ratio = 1.55; 95% confidence interval: 1.07, 2.24). In conclusion, we assessed the effectiveness of HCQ with or without azithromycin in treatment of patients hospitalized with COVID-19, using a national sample of the US veteran population. Using rigorous study design and analytic methods to reduce confounding and bias, we found no evidence of a survival benefit from the administration of HCQ.

Deep active learning for classifying cancer pathology reports.

BACKGROUND: Automated text classification has many important applications in the clinical setting; however, obtaining labelled data for training machine learning and deep learning models is often difficult and expensive. Active learning techniques may mitigate this challenge by reducing the amount of labelled data required to effectively train a model. In this study, we analyze the effectiveness of 11 active learning algorithms on classifying subsite and histology from cancer pathology reports using a Convolutional Neural Network as the text classification model. RESULTS: We compare the performance of each active learning strategy using two differently sized datasets and two different classification tasks. Our results show that on all tasks and dataset sizes, all active learning strategies except diversity-sampling strategies outperformed random sampling, i.e., no active learning. On our large dataset (15K initial labelled samples, adding 15K additional labelled samples each iteration of active learning), there was no clear winner between the different active learning strategies. On our small dataset (1K initial labelled samples, adding 1K additional labelled samples each iteration of active learning), marginal and ratio uncertainty sampling performed better than all other active learning techniques. We found that compared to random sampling, active learning strongly helps performance on rare classes by focusing on underrepresented classes. CONCLUSIONS: Active learning can save annotation cost by helping human annotators efficiently and intelligently select which samples to label. Our results show that a dataset constructed using effective active learning techniques requires less than half the amount of labelled data to achieve the same performance as a dataset constructed using random sampling.

COVID-19 Evidence Accelerator: A parallel analysis to describe the use of Hydroxychloroquine with or without Azithromycin among hospitalized COVID-19 patients.

BACKGROUND: The COVID-19 pandemic remains a significant global threat. However, despite urgent need, there remains uncertainty surrounding best practices for pharmaceutical interventions to treat COVID-19. In particular, conflicting evidence has emerged surrounding the use of hydroxychloroquine and azithromycin, alone or in combination, for COVID-19. The COVID-19 Evidence Accelerator convened by the Reagan-Udall Foundation for the FDA, in collaboration with Friends of Cancer Research, assembled experts from the health systems research, regulatory science, data science, and epidemiology to participate in a large parallel analysis of different data sets to further explore the effectiveness of these treatments. METHODS: Electronic health record (EHR) and claims data were extracted from seven separate databases. Parallel analyses were undertaken on data extracted from each source. Each analysis examined time to mortality in hospitalized patients treated with hydroxychloroquine, azithromycin, and the two in combination as compared to patients not treated with either drug. Cox proportional hazards models were used, and propensity score methods were undertaken to adjust for confounding. Frequencies of adverse events in each treatment group were also examined. RESULTS: Neither hydroxychloroquine nor azithromycin, alone or in combination, were significantly associated with time to mortality among hospitalized COVID-19 patients. No treatment groups appeared to have an elevated risk of adverse events. CONCLUSION: Administration of hydroxychloroquine, azithromycin, and their combination appeared to have no effect on time to mortality in hospitalized COVID-19 patients. Continued research is needed to clarify best practices surrounding treatment of COVID-19.

Limitations of Transformers on Clinical Text Classification.

Bidirectional Encoder Representations from Transformers (BERT) and BERT-based approaches are the current state-of-the-art in many natural language processing (NLP) tasks; however, their application to document classification on long clinical texts is limited. In this work, we introduce four methods to scale BERT, which by default can only handle input sequences up to approximately 400 words long, to perform document classification on clinical texts several thousand words long. We compare these methods against two much simpler architectures - a word-level convolutional neural network and a hierarchical self-attention network - and show that BERT often cannot beat these simpler baselines when classifying MIMIC-III discharge summaries and SEER cancer pathology reports. In our analysis, we show that two key components of BERT - pretraining and WordPiece tokenization - may actually be inhibiting BERT's performance on clinical text classification tasks where the input document is several thousand words long and where correctly identifying labels may depend more on identifying a few key words or phrases rather than understanding the contextual meaning of sequences of text.

Privacy-Preserving Deep Learning NLP Models for Cancer Registries.

Population cancer registries can benefit from Deep Learning (DL) to automatically extract cancer characteristics from the high volume of unstructured pathology text reports they process annually. The success of DL to tackle this and other real-world problems is proportional to the availability of large labeled datasets for model training. Although collaboration among cancer registries is essential to fully exploit the promise of DL, privacy and confidentiality concerns are main obstacles for data sharing across cancer registries. Moreover, DL for natural language processing (NLP) requires sharing a vocabulary dictionary for the embedding layer which may contain patient identifiers. Thus, even distributing the trained models across cancer registries causes a privacy violation issue. In this paper, we propose DL NLP model distribution via privacy-preserving transfer learning approaches without sharing sensitive data. These approaches are used to distribute a multitask convolutional neural network (MT-CNN) NLP model among cancer registries. The model is trained to extract six key cancer characteristics - tumor site, subsite, laterality, behavior, histology, and grade - from cancer pathology reports. Using 410,064 pathology documents from two cancer registries, we compare our proposed approach to conventional transfer learning without privacy-preserving, single-registry models, and a model trained on centrally hosted data. The results show that transfer learning approaches including data sharing and model distribution outperform significantly the single-registry model. In addition, the best performing privacy-preserving model distribution approach achieves statistically indistinguishable average micro- and macro-F1 scores across all extraction tasks (0.823,0.580) as compared to the centralized model (0.827,0.585).

Accelerated training of bootstrap aggregation-based deep information extraction systems from cancer pathology reports.

OBJECTIVE: In machine learning, it is evident that the classification of the task performance increases if bootstrap aggregation (bagging) is applied. However, the bagging of deep neural networks takes tremendous amounts of computational resources and training time. The research question that we aimed to answer in this research is whether we could achieve higher task performance scores and accelerate the training by dividing a problem into sub-problems. MATERIALS AND METHODS: The data used in this study consist of free text from electronic cancer pathology reports. We applied bagging and partitioned data training using Multi-Task Convolutional Neural Network (MT-CNN) and Multi-Task Hierarchical Convolutional Attention Network (MT-HCAN) classifiers. We split a big problem into 20 sub-problems, resampled the training cases 2,000 times, and trained the deep learning model for each bootstrap sample and each sub-problem-thus, generating up to 40,000 models. We performed the training of many models concurrently in a high-performance computing environment at Oak Ridge National Laboratory (ORNL). RESULTS: We demonstrated that aggregation of the models improves task performance compared with the single-model approach, which is consistent with other research studies; and we demonstrated that the two proposed partitioned bagging methods achieved higher classification accuracy scores on four tasks. Notably, the improvements were significant for the extraction of cancer histology data, which had more than 500 class labels in the task; these results show that data partition may alleviate the complexity of the task. On the contrary, the methods did not achieve superior scores for the tasks of site and subsite classification. Intrinsically, since data partitioning was based on the primary cancer site, the accuracy depended on the determination of the partitions, which needs further investigation and improvement. CONCLUSION: Results in this research demonstrate that 1. The data partitioning and bagging strategy achieved higher performance scores. 2. We achieved faster training leveraged by the high-performance Summit supercomputer at ORNL.

Knowledge Graph-Enabled Cancer Data Analytics.

Cancer registries collect unstructured and structured cancer data for surveillance purposes which provide important insights regarding cancer characteristics, treatments, and outcomes. Cancer registry data typically (1) categorize each reportable cancer case or tumor at the time of diagnosis, (2) contain demographic information about the patient such as age, gender, and location at time of diagnosis, (3) include planned and completed primary treatment information, and (4) may contain survival outcomes. As structured data is being extracted from various unstructured sources, such as pathology reports, radiology reports, medical records, and stored for reporting and other needs, the associated information representing a reportable cancer is constantly expanding and evolving. While some popular analytic approaches including SEER*Stat and SAS exist, we provide a knowledge graph approach to organizing cancer registry data. Our approach offers unique advantages for timely data analysis and presentation and visualization of valuable information. This knowledge graph approach semantically enriches the data, and easily enables linking with third-party data which can help explain variation in cancer incidence patterns, disparities, and outcomes. We developed a prototype knowledge graph based on the Louisiana Tumor Registry dataset. We present the advantages of the knowledge graph approach by examining: i) scenario-specific queries, ii) links with openly available external datasets, iii) schema evolution for iterative analysis, and iv) data visualization. Our results demonstrate that this graph based solution can perform complex queries, improve query run-time performance by up to 76%, and more easily conduct iterative analyses to enhance researchers' understanding of cancer registry data.

Using case-level context to classify cancer pathology reports.

Individual electronic health records (EHRs) and clinical reports are often part of a larger sequence-for example, a single patient may generate multiple reports over the trajectory of a disease. In applications such as cancer pathology reports, it is necessary not only to extract information from individual reports, but also to capture aggregate information regarding the entire cancer case based off case-level context from all reports in the sequence. In this paper, we introduce a simple modular add-on for capturing case-level context that is designed to be compatible with most existing deep learning architectures for text classification on individual reports. We test our approach on a corpus of 431,433 cancer pathology reports, and we show that incorporating case-level context significantly boosts classification accuracy across six classification tasks-site, subsite, laterality, histology, behavior, and grade. We expect that with minimal modifications, our add-on can be applied towards a wide range of other clinical text-based tasks.