FORMULA: FactORized MUlti-task LeArning for task discovery in personalized medical models

By Jianpeng Xu, Jiayu Zhou, Pang-Ning Tan. 2015

33 citations recorded by the source corpus.

Medical predictive modeling is a challenging problem due to the heterogeneous nature of the patients. In order to build effective medical predictive models we need to address such heterogeneous nature during modeling and allow patients to have their own personalized models instead of using a one-size-fits-all model. However, building a personalized model for each patient is computationally expensive and the over-parametrization of the model makes it susceptible to the model overfitting problem. To address these challenges, we propose a novel approach called FactORized MUlti-task LeArning model (Formula), which learns the personalized model of each patient via a sparse multi-task learning method. The personalized models are assumed to share a low-rank representation, known as the base models. Formula is designed to simultaneously learn the base models as well as the personalized model of each patient, where the latter is a linear combination of the base models. We have performed extensive experiments to evaluate the proposed approach on a real medical data set. The proposed approach delivered superior predictive performance while the personalized models offered many useful medical insights.

Subfields: Artificial Intelligence, Health Information Management, Psychiatry and Mental health

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Qiang Yang
2. Yu Zhang
3. Harini Suresh
4. John V. Guttag
5. Jiayu Zhou
6. Jen J. Gong
7. Jianpeng Xu
8. Lifeng Luo

In Health Information Management

1. Harini Suresh
2. John V. Guttag
3. Qiang Yang
4. Yu Zhang
5. Jen J. Gong

In Psychiatry and Mental health

1. Qiang Yang
2. Yu Zhang
3. John V. Guttag
4. Jen J. Gong
5. Harini Suresh
6. Jianpeng Xu
7. Pang-Ning Tan
8. Lifeng Luo

Citations in the corpus

• A Survey on Multi-Task Learning (2017, methodology): “In [192], a task clustering method is proposed to learn a personalized medical model by factorizing the parameter”
• Learning Tasks for Multitask Learning: Heterogenous Patient Populations in the ICU (2018, methodology): “Task formulations for multi-task learning with clinical data fall into two categories: 1) distinct outcomes are used as tasks [4, 18, 34, 42] and 2) distinct patient populations are used as tasks [30, 46].”
• Online Multi-Task Learning Framework for Ensemble Forecasting (2017, background): “[30] presented a formulation that combines both low-rank representation and explicit task relations for developing personalized medical models of patients.”
• GSpartan: a Geospatio-Temporal Multi-task Learning Framework for Multi-location Prediction (2016, background): “The framework also enables domain-specific constraints such as spatial autocorrelation to be integrated into its formulation.”