Shan Xu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Shan Xu has a total of 28 co-authors. Here are the top co-authors by number of shared publications:
Xiaohui Duan
School of Electronic Engineering and Computer Science, Peking University, Beijing, China
3 shared publicationsDong Min
2 shared publicationsTiangang Zhu
Peking University People's Hospital, Beijing, China
2 shared publicationsAnpeng Huang
Lab of Mobile Health, Peking University, Beijing, China
1 shared publicationBingli Jiao
Lab of Mobile Health, Peking University, Beijing, China
1 shared publicationBotong Liu
Lab of Mobile Health, Peking University, Beijing, China
1 shared publication
Latest Publications
- Open Access
- 1 September 2021
- Published by Elsevier in Intelligent Medicine
- Vol. 1 (3) , 134-136
Digital inclusion of older people: harnessing digital technologies to promote healthy ageing in the Western Pacific Region
- 1 September 2021
- Published by Elsevier in Intelligent Medicine
- Vol. 1 (3) , 134-136
Open AccessDigital inclusion of older people: harnessing digital technologies to promote healthy ageing in the Western Pacific Region
- 1 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-10
Preparing for the Ai Era Under the Digital Health Framework
- 1 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-10
Preparing for the Ai Era Under the Digital Health Framework
- 1 July 2019
- Published by Elsevier in The Lancet
- Vol. 394 (10192) , 9-11
WHO and ITU establish benchmarking process for artificial intelligence in health
- 1 July 2019
- Published by Elsevier in The Lancet
- Vol. 394 (10192) , 9-11
WHO and ITU establish benchmarking process for artificial intelligence in health
- 1 March 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 228-232
Cardiovascular risk prediction method based on CFS subset evaluation and random forest classification framework
Cardiovascular Disease (CVD) is a highly significant contributor to loss of quality and quantity of life all over the world. Early detection and risk prediction is very important for patients' treatment and doctors' diagnose. This paper focus on establishing a more accurate and practical risk prediction system based on data mining techniques to provide auxiliary medical service. In order to be practically used for collecting and analyzing patients' data in healthcare industries, the system consists of four parts: data interface, data preparation, feature selection and classification. Data interface response to obtain hospitals' raw data from hospital; data preprocessing is needed for data integration, data cleaning and rating mapping etc. Key features were then selected by CFS Subset Evaluation combined with Best-First-Search method to reduce dimensionality. Random forest was inducted as basic classifier to identify risk level, which is a prior trial in CVD risk prediction field. Cleveland Heart-Disease Database (CHDD) and Cardiology inpatient dataset of PKU People's Hospital were both tested to confirm accuracy as well as practicality. In CHDD test, our system has a significantly higher accuracy of 91.6% than other methods. In People's Hospital dataset test, it achieves an accuracy of 97%, which is better than most of other classifiers except SVM (98.9%), however random forest only take half of time than SVM. Comprehensively considering the risk prediction system shows great significance in accuracy and practical use for patients' treatment and doctors' diagnose.
- 1 March 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 228-232
Cardiovascular risk prediction method based on CFS subset evaluation and random forest classification framework
Cardiovascular Disease (CVD) is a highly significant contributor to loss of quality and quantity of life all over the world. Early detection and risk prediction is very important for patients' treatment and doctors' diagnose. This paper focus on establishing a more accurate and practical risk prediction system based on data mining techniques to provide auxiliary medical service. In order to be practically used for collecting and analyzing patients' data in healthcare industries, the system consists of four parts: data interface, data preparation, feature selection and classification. Data interface response to obtain hospitals' raw data from hospital; data preprocessing is needed for data integration, data cleaning and rating mapping etc. Key features were then selected by CFS Subset Evaluation combined with Best-First-Search method to reduce dimensionality. Random forest was inducted as basic classifier to identify risk level, which is a prior trial in CVD risk prediction field. Cleveland Heart-Disease Database (CHDD) and Cardiology inpatient dataset of PKU People's Hospital were both tested to confirm accuracy as well as practicality. In CHDD test, our system has a significantly higher accuracy of 91.6% than other methods. In People's Hospital dataset test, it achieves an accuracy of 97%, which is better than most of other classifiers except SVM (98.9%), however random forest only take half of time than SVM. Comprehensively considering the risk prediction system shows great significance in accuracy and practical use for patients' treatment and doctors' diagnose.
- 1 March 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 7, 1-5
Cardiovascular risk prediction method based on test analysis and data mining ensemble system
Cardiovascular disease (CVD) is a highly significant contributor to loss of quality and quantity of life all over the world. Early detection and prediction is very important for patients' treatment and doctors' diagnose which can help to reduce mortality. In this paper, we focus on practical problem of Chinese hospital dealing with cardiovascular patients' data to make an early detection and risk prediction. To better understand the prescription and advice in Chinese, basic natural language processing method was used to synonym recognition and attribute extraction in Ultrasonic echocardiography. After data preprocessing, over 50 data mining techniques was tested for real patents dataset. Totake full advantage of multi-methods and reduce bias, top 6 subclassifiers was selected to form an ensemble system, adjusted voting mechanism was used to make a final result, which consists of risk prediction and confidence. System has a high precision of 79.3% for 2628 cases of real patents in experiment. Therisk prediction confidence and algorithm accuracy shows great significance in practical use for doctors' diagnosing.
- 1 March 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 7, 1-5
Cardiovascular risk prediction method based on test analysis and data mining ensemble system
Cardiovascular disease (CVD) is a highly significant contributor to loss of quality and quantity of life all over the world. Early detection and prediction is very important for patients' treatment and doctors' diagnose which can help to reduce mortality. In this paper, we focus on practical problem of Chinese hospital dealing with cardiovascular patients' data to make an early detection and risk prediction. To better understand the prescription and advice in Chinese, basic natural language processing method was used to synonym recognition and attribute extraction in Ultrasonic echocardiography. After data preprocessing, over 50 data mining techniques was tested for real patents dataset. Totake full advantage of multi-methods and reduce bias, top 6 subclassifiers was selected to form an ensemble system, adjusted voting mechanism was used to make a final result, which consists of risk prediction and confidence. System has a high precision of 79.3% for 2628 cases of real patents in experiment. Therisk prediction confidence and algorithm accuracy shows great significance in practical use for doctors' diagnosing.
- 1 August 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2015 (1094687X) , 2043-2046
Design and implementation of an intelligent belt system using accelerometer
Activity monitor systems are increasing used recently. They are important for athletes and casual users to manage physical activity during daily exercises. In this paper, we use a triaxial accelerometer to design and implement an intelligent belt system, which can detect the user's step and flapping motion. In our system, a wearable intelligent belt is worn on the user's waist to collect activity acceleration signals. We present a step detection algorithm to detect real-time human step, which has high accuracy and low complexity. In our system, an Android App is developed to manage the intelligent belt. We also propose a protocol, which can guarantee data transmission between smartphones and wearable belt effectively and efficiently. In addition, when users flap the belt in emergency, the smartphone will receive alarm signal sending by the belt, and then notifies the emergency contact person, which can be really helpful for users in danger. Our experiment results show our system can detect physical activities with high accuracy (overall accuracy of our algorithm is above 95%) and has an effective alarm subsystem, which is significant for the practical use.
- 1 August 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2015 (1094687X) , 2043-2046
Design and implementation of an intelligent belt system using accelerometer
Activity monitor systems are increasing used recently. They are important for athletes and casual users to manage physical activity during daily exercises. In this paper, we use a triaxial accelerometer to design and implement an intelligent belt system, which can detect the user's step and flapping motion. In our system, a wearable intelligent belt is worn on the user's waist to collect activity acceleration signals. We present a step detection algorithm to detect real-time human step, which has high accuracy and low complexity. In our system, an Android App is developed to manage the intelligent belt. We also propose a protocol, which can guarantee data transmission between smartphones and wearable belt effectively and efficiently. In addition, when users flap the belt in emergency, the smartphone will receive alarm signal sending by the belt, and then notifies the emergency contact person, which can be really helpful for users in danger. Our experiment results show our system can detect physical activities with high accuracy (overall accuracy of our algorithm is above 95%) and has an effective alarm subsystem, which is significant for the practical use.
- 1 May 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 605-609
A combination of new static detection and dynamic detection in people counting
With the improvement of the intelligence degree of video surveillance system, the people-counting in surveillance area becomes a studying focus. This paper proposes a fast and effective people-counting method. In this method, we divide the monitoring area into blocks at first, and then we recognize people in each block and combine both the static detection and the dynamic detection. In the static detection part, we use covariance algorithm to avoid factitious adjustment of threshold. As a result, it will be more persuasive by comparing the entire difference distributions. What is more, some interference of non-human objects can be excluded effectively by using the dynamic detection. Therefore we can get a more accurate final result.
- 1 May 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 605-609
A combination of new static detection and dynamic detection in people counting
With the improvement of the intelligence degree of video surveillance system, the people-counting in surveillance area becomes a studying focus. This paper proposes a fast and effective people-counting method. In this method, we divide the monitoring area into blocks at first, and then we recognize people in each block and combine both the static detection and the dynamic detection. In the static detection part, we use covariance algorithm to avoid factitious adjustment of threshold. As a result, it will be more persuasive by comparing the entire difference distributions. What is more, some interference of non-human objects can be excluded effectively by using the dynamic detection. Therefore we can get a more accurate final result.