Large scale characterization for health equity assessment

Abstract: Health disparities in the United States are one of the largest factors in reducing the health of the population. Disparities means some groups have lower life expectancy, are dying at higher rates from COVID-19, and utilize less mental health services, to name just a few examples. The future of medicine will be based on Artificial Intelligence and new technological platforms that promise to improve outcomes and reduce cost. Our role as AI researchers should be to ensure that these new technologies also reduce health disparities. In this talk I will describe recent work showing how we can work to reduce health disparities in the future of medicine. By ensuring that our task, datasets, algorithms and evaluations are equitable and representative of all types of patients, we can ensure that the research we develop will reduce health disparities. **Bio**: Mark Dredze is the John C Malone Associate Professor of Computer Science at Johns Hopkins University. He is affiliated with the Malone Center for Engineering in Healthcare, the Center for Language and Speech Processing, among others. He holds a secondary appointment in the Department of Health Sciences Informatics in the School of Medicine. He obtained his PhD from the University of Pennsylvania in 2009. Prof. Dredze’s research develops statistical models of language with applications to social media analysis, public health and clinical informatics. Within Natural Language Processing he focuses on statistical methods for information extraction but has considered a wide range of NLP tasks, including syntax, semantics, sentiment and spoke language processing. His work in public health includes tobacco control, vaccination, infectious disease surveillance, mental health, drug use, and gun violence prevention. He also develops new methods for clinical NLP on medical records.

Please join us in [Gather.town](https://gather.town/app/OVLlsG1Ar09XPYTN/ML4H2020) for our virtual poster session. Below are the poster titles with their poster numbers to the left. 1. Trust Issues - Uncertainty Estimation Does not Enable Reliable OOD Prediction On Medical Tabular Data 2. Improved Clinical Abbreviation Expansion via Non-Sense-Based Approaches 3. Concept-based model explanations for Electronic Health Records 4. DeepHeartBeat: Sequence modelling for medical data 5. Temporal Pointwise Convolution Networks for Length of Stay Prediction in the ICU 6. Parkinsonian Chinese Speech Analysis towards Automatic Classification of Parkinson's Disease 7. Evaluation of Contrastive Predictive Coding for Histopathology Applications 8. Multiomics Data Analysis Predicts Risk of Preeclampsia 9. Confounding Feature Acquisition for Causal Effect Estimation 10. ParaMed: A Parallel Corpus for English-Chinese Translation in the Biomedical Domain 11. Natural Language Processing to Detect Cognitive Concerns in Electronic Health Records Using Deep Learning 12. Mobility network models of COVID-19 explain inequities and inform reopening 13. Incorporating Healthcare Motivated Constraints in Restless Multi-Armed Bandit Based Resource Allocation 14. Zero-Shot Clinical Acronym Expansion via Latent Meaning Cells 15. Decomposing Normal and Abnormal Features of Medical Images for Content-based Image Retrieval 16. Cost-Sensitive Machine Learning Classification for Mass Tuberculosis Verbal Screening 17. Uncertainty-Aware Counterfactual Explanations for Medical Diagnosis 18. Detecting small polyps using a Dynamic SSD-GAN 19. The unreasonable effectiveness of Batch-Norm statistics in addressing catastrophic forgetting across medical institutions 20. Learning transition times in event sequences: the Event-Based Hidden Markov Model of disease progression 21. Neural Temporal Point Processes For Modelling Electronic Health Records 22. Comparison of pharmacist evaluation of medication orders with predictions of a machine learning model 23. Medical symptom recognition from patient text: An active learning approach for long-tailed multilabel distributions 24. Accounting for Affect in Pain Level Recognition 25. Forecasting Emergency Department Capacity Constraints for COVID Isolation Beds 26. Spectral discontinuity design: Interrupted time series with spectral mixture kernels 27. A Bayesian Approach for Continual Learning in Clinical Time Series 28. A Neural SIR Model for Global Forecasting 29. MTB-HINE-BERT: a pre-trained genetic mutation representation model for predicting drug resistance of Mycobacterium tuberculosis 30. Stable predictions for health related anticausal prediction tasks affected by selection biases: the need to deconfound the test set features 31. Deep Learning Derived Histopathology Image Score for Increasing Phase 3 Clinical Trial Probability of Success 32. Self-Supervision Closes the Gap Between Weak and Strong Supervision in Histology 33. Model-Attentive Ensemble Learning for Sequence Modeling 34. Towards Automated Anamnesis Summarization: BERT-based Models for Symptom Extraction 35. Recommendations for Bayesian hierarchical model specifications for case-control studies in mental health 36. A decision-making tool to fine-tune abnormal levels in the complete blood count tests 37. Enhancing COVID-19 patient deterioration prediction using secondary task pre-trained embedding 38. Stratification of Systemic Lupus Erythematosus Patients with Gene Expression Data to Reveal Expression of Distinct Immune Pathways 39. Phenotyping Clusters of Patient Trajectories suffering from Complex Disease 40. Image analysis for Alzheimer's disease prediction: Embracing pathological hallmarks for model architecture design 41. Deep Cox Mixtures for Survival Regression 41

The year 2020 has brought into focus a second pandemic of social injustice and systemic bias with the disproportionate deaths observed for minority patients infected with COVID. As we observe an increase in development and adoption of AI for medical care, we note variable performance of the models when tested on previously unseen datasets, and also bias when the outcome proxies such as healthcare costs are utilized. Despite progressive maturity in AI development with increased availability of large open source datasets and regulatory guidelines, operationalizing fairness is difficult and remains largely unexplored. In this talk, we review the background/context for FAIR and UNFAIR sequelae of AI algorithms in healthcare, describe practical approaches to FAIR Medical AI, and issue a grand challenge with open/unanswered questions. **Bio**: Dr. Gichoya is a multidisciplinary researcher, trained as both an informatician and a clinically active radiologist. She is an assistant professor at Emory university, and works in Interventional Radiology and Informatics. She has been funded through the Grand Challenges Canada, NBIB and NSF ECCS. Her career focus is on validating machine learning models for health in real clinical settings, exploring explainability, fairness, and a specific focus on how algorithms fail. She has worked on the curation of datasets for the SIIM (Society for Imaging Informatics in Medicine) hackathon and ML committee. She volunteers on the ACR and RSNA machine learning committees to support the AI ecosystem to advance development and use of AI in medicine. She is currently working on the sociotechnical context for AI explainability for radiology, especially the dimensions of human factors that govern user perceptions and preferences of XAI systems.

Please join us in [Gather.town](https://gather.town/app/OVLlsG1Ar09XPYTN/ML4H2020) for our virtual poster session. Below are the poster titles with their poster numbers to the left. 42\. Towards Diagnosing Nonalcoholic Fatty Liver Disease with Abdominal MRI Data using Deep Learning 43\. Adversarial Factor Models for Confound Disentangled Autism Biomarkers 44\. Identifying Decision Points for Safe and Interpretable RL in Hypotension Treatment 45\. Neural ODEs for Multi-State Survival Analysis 46\. Assessing racial inequality in COVID-19 testing with Bayesian threshold tests 47\. Learning to Predict and Support for Clinical Risk Stratification 48\. sEMG Gesture Recognition with a Simple Model of Attention 49\. Addressing the Real-world Class Imbalance Problem in Dermatology 50\. Appropriate Evaluation of Diagnostic Utility of Machine Learning Algorithm Generated Images 51\. Exploring Gender Disparities in Time to Diagnosis 52\. Contrastive Representation Learning for Electroencephalogram Classification 53\. Toward Understanding Clinical Context of Medication Change Events in Clinical Narratives 54\. Pneumothorax and chest tube classification on chest x-rays for detection of missed pneumothorax 55\. Interpretable Epilepsy Detection in Routine, Interictal EEG Data using Deep Learning 56\. ML4H Auditing: From Paper to Practice 57\. EEG-GCNN: Augmenting Electroencephalogram-based Neurological Disease Diagnosis using a Domain-guided Graph Convolutional Neural Network 58\. GloFlow: Global Image Alignment for Creation of Whole Slide Images for Pathology from Video 59\. A Bayesian Hierarchical Network for Combining Heterogeneous Data Sources in Medical Diagnoses 60\. An Interpretable End-to-end Fine-tuning Approach for Long Clinical Text 61\. A Study of Domain Generalization on Ultrasound-based Multi-Class Segmentation of Arteries, Veins, Ligaments, and Nerves Using Transfer Learning 62\. CliniQG4QA: Generating Diverse Questions for Domain Adaptation of Clinical Question Answering 63\. TL-Lite: Visualization and Temporal Learning for Clinical Cohorts 64\. Estimating County-Level COVID-19 Exponential Growth Rates Using Generalized Random Forests 65\. CheXphoto: 10,000+ Photos and Transformations of Chest X-rays for Benchmarking Deep Learning Robustness 66\. Augmenting BERT Carefully with Underrepresented Linguistic Features 67\. Towards Trainable Saliency Maps in Medical Imaging 68\. transferGWAS: GWAS of Images Using Deep Transfer Learning 69\. CheXphotogenic: Generalization of Deep Learning Models for Chest X-ray Interpretation to Photos of Chest X-rays 70\. Generating SOAP Notes from Doctor-Patient Conversations 71\. Personalized Healthcare and Causal Interventions 72\. Attend and Decode: 4D fMRI Task State Decoding Using Attention Models 73\. An Empirical Study of Representation Learning for Reinforcement Learning in Healthcare 74\. Opening a Can of Words: Train-test overlaps in Clinical Natural Language Processing Datasets 75\. A stability-driven protocol for drug response interpretable prediction (staDRIP) 76\. Quantifying Common Support between Multiple Treatment Groups Using a Contrastive VAE 77\. Learning Optimal Predictive Checklists 78\. 3D Photography Based Neural Network Craniosynostosis Triaging System 79\. A Time-Aware Transformer Based Model for Suicide Ideation Detection on Social Media 80\. Bayesian Recurrent State Space Model For rs-fMRI 81\. COVID-19 in Differential Diagnosis of Online Symptom Assessments 82\. Uncertainty-Aware Multi-Modal Ensembling for Severity Prediction of Alzheimer’s Dementia

Abstract: Recent advances in training deep learning algorithms have demonstrated potential to accommodate the complex variations present in medical data. In this talk, I will describe technical advancements and challenges in the development and clinical application of deep learning algorithms designed to interpret medical images. I will also describe advances and current challenges in the deployment of medical imaging deep learning algorithms into practice. This talk presents work that is jointly done with Matt Lungren, Curt Langlotz, Nigam Shah, and several more collaborators. **Bio**: Andrew Ng is Founder of [DeepLearning.AI](https://www.deeplearning.ai/), Founder and CEO of Landing AI, General Partner at AI Fund, Chairman and Co-Founder of Coursera, and an Adjunct Professor at Stanford University. As a pioneer both in machine learning and online education, Dr. Ng has changed countless lives through his work and research in the field of artificial intelligence.

Abstract: Recent advances in training deep learning algorithms have demonstrated potential to accommodate the complex variations present in medical data. In this talk, I will describe technical advancements and challenges in the development and clinical application of deep learning algorithms designed to interpret medical images. I will also describe advances and current challenges in the deployment of medical imaging deep learning algorithms into practice. This talk presents work that is jointly done with Matt Lungren, Curt Langlotz, Nigam Shah, and several more collaborators. **Bio**: Andrew Ng is Founder of Founder and CEO of Landing AI, General Partner at AI Fund, Chairman and Co-Founder of Coursera, and an Adjunct Professor at Stanford University. As a pioneer both in machine learning and online education, Dr. Ng has changed countless lives through his work and research in the field of artificial intelligence.