Tehran University of Medical Sciences
Office of Vice-Chancellor for Global Strategies & International Affairs
International Human Capacity Development (IHCD)
Code : 9822-353503      Publish Date : Saturday, October 14, 2017 Visit : 2311

Intl. Congress form | International Congress Report | International Congress Report For Students and Staff | The 39th Conference on Stochastic Processes and their Applications (SPA2017) by Farzan Madadizadeh

The 39th Conference on Stochastic Processes and their Applications (SPA2017) by Farzan Madadizadeh
The 39th Conference on Stochastic Processes and their Applications (SPA2017) by Farzan Madadizadeh
Farzan Madadizadeh | International Congress Form

Application Code :
762-0217-0062
 
Created Date : Tuesday, October 3, 2017-23:28 23:28:17Update Date : Wednesday, October 4, 2017-11:39 11:39:57
IP Address : 194.225.61.124Submit Date : Wednesday, October 4, 2017-11:40 11:40:55Email : fmadadizadeh@razi.tums.ac.ir
Personal Information
Name : Farzan
Surname : Madadizadeh
School/Research center : School of Public Health
If you choose other, please name your Research center :  
Position : student
Tel : +98-913-2475024
Information of Congress
Title of the Congress : The 39th Conference on Stochastic
Processes and their Applications (SPA2017)
Title of your Abstract : A new approach in classification discrete survival data by discrete hidden Markov model and comparison to one common method
Destination Country : Russia
From : Monday, July 24, 2017
To : Friday, July 28, 2017
Abstract(Please copy/paste the abstract send to the congress) : Background: Diagnosis of cancer survival is very important and existence of reliable system is necessary for reduce the medical expenditure. Discrete Hidden Markov Models (DHMMs) are a ubiquitous tools and probabilistic techniques for modeling sequence data. These models include an underlying stochastic process that is hidden but could be inferred through the observations it generates. Discrete time Survival data comprised of a set of events and observation during the discrete time.
This paper presents the performance of DHMM in classification of event and compere results with logistic regression that considers overall probability of occurrence of event.

Methods: A dataset was acquired of Health Department and Cancer registry of Kerman Province which is located in south of Iran, and data includes the information of 900 breast cancer patients aged 15 to 80 years, and its seven associated risk factors among patients since 1999 to 2007. Our breast cancer data are assumed to include two possible events such as live and died. DHMM was trained based on a set of 675 patients information (75% of data) and that was validate in a test set of 225 patients information (25% of data). The Area Under the ROC Curve (AUC), sensitivity, specificity and accuracy used as measures of validate of the efficiency of model in prediction of patients survival status.

Results: Sensitivity, specificity, accuracy and the and area under the ROC curve of the logistic regression was 0.86, 0.97,0.92 and 0.951, respectively. The sensitivity, specificity, accuracy and the area under the ROC curve of DHMM was 0.989, 0.99, 0.939 and 0.94, respectively.

Conclusions: According to the four evaluation criteria, DHMM would give better performance than logistic regression.

Acknowledgement: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.


References
[1] Fink GA (2014). Hidden Markov Models. Markov Models for Pattern Recognition, Springer, 71-106.

[2] Rabiner L (1989). A tutorial on hidden Markov models and selected applications in speech recognition, Proceedings of the IEEE, 77(2), 257-86.

[3] Hassan MR, Hossain MM, Begg RK, Ramamohanarao K, Morsi Y (2010). Breast-cancer identification using HMM-fuzzy approach, Computers in biology and medicine, 40(3), 240-51.
Keywords of your Abstract : discrete survival data analysis, discrete hidden Markov model
Acceptance Letter : http://gsia.tums.ac.ir/images/UserFiles/41093/Forms/762/acceptance_letter.pdf
The presentation : Oral
The Cover of Abstract book : http://gsia.tums.ac.ir/images/UserFiles/41093/Forms/762/cover.pdf
Published abstract in the abstract book with the related code : http://gsia.tums.ac.ir/images/UserFiles/41093/Forms/762/my_abstract_in_conference_book.pdf
Where has your abstract been indexed? : other
If you choose other, please name : Congress Site
The Congress Reporting Form
How many volunteers were present at the Congress? : more than 1500
Delegates from which countries presented in the congress? : USA, Spain, France, China, Italy, Russia, Sweden,Finland, Portugal, Japan, Austria, Germany, UK, Poland, Serbia, Australia,
Were the delegates of any other organizations present in the congress? : Yes
If yes, please write the names of the organizations in the box : Amir Kabir University Of Iran
What were the responses to your talking points? Were specific questions or concerns raised? : They said It is very good and new issue in modeling survival data.
If you met staff members, please list their full names & positions. : Vladimir Spokoiny, Berlin, Germany
Maxim Chuyashkin, Moscow, Russia
I don not know anyone else.
Please inform us if there are any follow up actions we need to talk with the members of the congress : No there is not any need for following up action to talk with the members of the congress because all steps were fully organized and they kindly answer the all question and have totally cooperation even after finishing the SPA congress, if there is any request or question the reply is ready as soon as possible.
Your experiences about the travel processes(Providing ticket, accommodation,...) : You must register in conference and pay amount of registration fee as soon as possible.
There is a big problem with get Visa to abroad.
Please give a briefing of your own observations and outcomes of the congress: : SPA is the largest statistical Congress which is held every year, scientists, students and researchers from all over the world participate in this event and their most recent finding in this field are presented there. It is unbelievable speed in knowledge growth specially in Markov model applications.
As my interest is hidden Markov models and random process, I chose these session in all 4 day program. It is noticeable all these topics were so crowded and sometime you should stay to listen the lecture. New method in Medical research, as well as new approaches, new modeling are in challenge and some are in use.
Beside the congress was exhibition hall and new tutorial book were introduced