Aaai2022 latest "time series data processing" report, 127 pages of PPT describing time series data processing and medical application progress

International Conference on artificial intelligence AAAI 2022 The paper It will be held online all the way , Time in 2 month 22 Japan -3 month 1 Japan , This session of the General Assembly will also be the... Th 36 the AAAI The conference . The conference covered many recent studies Tutorial The report , come from Mihaela van der Schaar And other scholars jointly made a progress report on medical and health time series data processing , Very noteworthy !

Time series data set , Such as electronic health records (EHR) And the registration form , Represents a valuable that spans the patient's entire life cycle ( But not perfect. ) Information sources for . Learning from time data is a mature field , It has been introduced in some previous tutorials , The medical field presents unique problems and challenges , New methods and ways of thinking are needed . Perhaps the most common application of time series is prediction . Although we will discuss the most advanced methods of disease prediction , But we will also focus on other important issues in time series , Such as event time or survival analysis 、 Personalized monitoring and treatment effects over time . These topics will be introduced in the context of healthcare , But they have wide applicability to other fields besides medicine . Besides , We will explore several essential features , Make artificial intelligence and machine learning models as useful as possible in clinical environments . We will discuss automatic machine learning , And solve the challenges of understanding and interpreting machine learning models and uncertainty estimation , Both in high-risk scenarios ( Like health care ) Are crucial . Our goal is to acquire the minimum necessary prerequisite knowledge . However , We will assume a basic knowledge of standard machine learning methods ( Such as MLPs、rnn). Although our tutorial will include some detailed explanations of machine learning techniques , But the focus will be on problem areas , Their unique challenges , And the way of thinking to overcome these problems .

The contents of the catalog ：

• Time series: the backbone of bespoke medicine

• The multi-faceted nature of time series

• Tailoring development of time series models to healthcare challenges

• Forecasting disease trajectories

• Time-to-event and survival analysis

• Clustering and phenotyping

• Screening and monitoring

• Early diagnosis

• Treatment effects

• Unlocking the full potential of time series models

• AutoML

• Interpretability

• Discovery and understanding of event data

• Uncertainty estimation

• (Informatively) missing data

• Synthetic data generation

• Reproducibility and visualization

   

Address ：

https://www.vanderschaar-lab.com/time-series-in-healthcare/