Learning Tasks for Multitask Learning

Suresh, Harini; Gong, Jen J.; Guttag, John V.

doi:10.1145/3219819.3219930

Cited by 41 publications

(12 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Jiang et al [ 37 ] used machine learning to predict in-hospital mortality in sepsis survivors (sepsis: AUROC 0.732; nonsepsis: AUROC 0.830). Suresh et al [ 38 ] developed a multitask model (AUROC 0.869) for mortality prediction that outperformed global and separate models. Fan et al [ 39 ] predicted in-hospital mortality for acute myocardial infarction patients by building several models such as logistic regression, decision tree, extreme gradient boosting, and random forest; among which, the logistic regression model performed best (AUROC 0.870).…”

Section: Discussionmentioning

confidence: 99%

Sequential Data–Based Patient Similarity Framework for Patient Outcome Prediction: Algorithm Development

Wang¹,

Wang²,

Zhou³

et al. 2022

J Med Internet Res

View full text Add to dashboard Cite

Background Sequential information in electronic medical records is valuable and helpful for patient outcome prediction but is rarely used for patient similarity measurement because of its unevenness, irregularity, and heterogeneity. Objective We aimed to develop a patient similarity framework for patient outcome prediction that makes use of sequential and cross-sectional information in electronic medical record systems. Methods Sequence similarity was calculated from timestamped event sequences using edit distance, and trend similarity was calculated from time series using dynamic time warping and Haar decomposition. We also extracted cross-sectional information, namely, demographic, laboratory test, and radiological report data, for additional similarity calculations. We validated the effectiveness of the framework by constructing k–nearest neighbors classifiers to predict mortality and readmission for acute myocardial infarction patients, using data from (1) a public data set and (2) a private data set, at 3 time points—at admission, on Day 7, and at discharge—to provide early warning patient outcomes. We also constructed state-of-the-art Euclidean-distance k–nearest neighbor, logistic regression, random forest, long short-term memory network, and recurrent neural network models, which were used for comparison. Results With all available information during a hospitalization episode, predictive models using the similarity model outperformed baseline models based on both public and private data sets. For mortality predictions, all models except for the logistic regression model showed improved performances over time. There were no such increasing trends in predictive performances for readmission predictions. The random forest and logistic regression models performed best for mortality and readmission predictions, respectively, when using information from the first week after admission. Conclusions For patient outcome predictions, the patient similarity framework facilitated sequential similarity calculations for uneven electronic medical record data and helped improve predictive performance.

show abstract

Section: Discussionmentioning

confidence: 99%

Sequential Data–Based Patient Similarity Framework for Patient Outcome Prediction: Algorithm Development

Wang¹,

Wang²,

Zhou³

et al. 2022

J Med Internet Res

View full text Add to dashboard Cite

show abstract

“…Evaluation bias occurs in the model design and/or building stage(s) during a model's evaluation and iteration when the model's parameters are biased. This is the result of poor model design governance and/or decision-making [36,66]. Again, this is "technical bias" as opposed to "social bias" [75], and 5.…”

Section: Model Outcome Bias: Post-hoc Explainability Methodsmentioning

confidence: 99%

Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law

2022

View full text Add to dashboard Cite

Organizations are increasingly employing complex black-box machine learning models in high-stakes decision-making. A popular approach to addressing the problem of opacity of black-box machine learning models is the use of post-hoc explainability methods. These methods approximate the logic of underlying machine learning models with the aim of explaining their internal workings, so that human examiners can understand them. In turn, it has been alluded that the insights from posthoc explainability methods can be used to help regulate black-box machine learning. This article examines the validity of these claims. By examining whether the insights derived from post-hoc explainability methods in post-model deployment can prima facie meet legal definitions in European (read European Union) non-discrimination law, we argue that machine learning post-hoc explanation methods cannot guarantee the insights they generate. Ultimately, we argue that the use of post-hoc explanatory methods is useful in many cases, but that these methods have limitations that prohibit reliance as the sole mechanism to guarantee fairness of model outcomes in high-stakes decision-making. By way of an ancillary function, the inadequacy of European Non-Discrimination Law for algorithmic decision-making is demonstrated too.

show abstract

“…Algorithmic bias is introduced during the MO-phase and results from inappropriate technical considerations. It can emerge when formulating the optimization problem, in which developers make data and parameters amenable to computers [18,21,42]. Resulting ML-models may fail to treat groups fairly under given conditions.…”

Section: Emergence Of Bias In Machine Learning Projectsmentioning

confidence: 99%

“…ML-models are often tested on the same benchmark to allow for an objective comparison. If the benchmark itself is not representative, models could be preferred that perform only well on a subset of the population [15,42].…”

Section: Emergence Of Bias In Machine Learning Projectsmentioning

confidence: 99%

“…Second, multitask learning is an efficient decoupling technique that can be added on top of black-box models to learn different classifiers for different groups, thereby mitigating algorithmic bias. It parametrizes different groups differently and learns simpler, multiple functions to account for group differences [42,62].…”

Section: Managing Bias In Machine Learning Projectsmentioning

confidence: 99%

See 1 more Smart Citation

Managing Bias in Machine Learning Projects

Fahse

Huber

Giffen

2021

Lecture Notes in Information Systems and Organisation

View full text Add to dashboard Cite

This paper introduces a framework for managing bias in machine learning (ML) projects. When ML-capabilities are used for decision making, they frequently affect the lives of many people. However, bias can lead to low model performance and misguided business decisions, resulting in fatal financial, social, and reputational impacts. This framework provides an overview of potential biases and corresponding mitigation methods for each phase of the wellestablished process model CRISP-DM. Eight distinct types of biases and 25 mitigation methods were identified through a literature review and allocated to six phases of the reference model in a synthesized way. Furthermore, some biases are mitigated in different phases as they occur. Our framework helps to create clarity in these multiple relationships, thus assisting project managers in avoiding biased ML-outcomes.

show abstract

Learning Tasks for Multitask Learning

Cited by 41 publications

References 39 publications

Sequential Data–Based Patient Similarity Framework for Patient Outcome Prediction: Algorithm Development

Sequential Data–Based Patient Similarity Framework for Patient Outcome Prediction: Algorithm Development

Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law

Managing Bias in Machine Learning Projects

Contact Info

Product

Resources

About