September, 10 — regular seminar
Event ended
Topic: “Evidence-based medicine and intelligent data analysis in clinical informatics using FCART system”
Speaker: Alexey Neznanov (PhD, Docent, Senior researcher at Laboratory for Intelligent Systems and Structural Analysis, NRU HSE)
The Laboratory for Comparative Social Research announces the next regular seminar, which will be held in Moscow (Myasnitskaya st., 22, room 511) on September 10 at 18-00 p.m. Alexey Neznanov (PhD, Docent, Senior researcher at Laboratory for Intelligent Systems and Structural Analysis, NRU HSE) will deliver a report “Evidence-based medicine and intelligent data analysis in clinical informatics using FCART system”.
Experiment design (or design of experiment (DOE), experiment protocol and experiment results analysis are the most important topics in any experimental work as well as formal base of scientific method. In the seminar we will consider the problem of checking quality of conducting statistical experiment and knowledge discovery from the statistical experiment data.
Systematic approach to statistical experiment planning and results analysis are present in small number of application domains. There are some illustrative examples of such domains.
1. Modern medicine with a conception of "Evidence-based medicine", that is the conscientious use of current best evidence in making decisions about the care of individual patients or the delivery of health services. Such best evidence is based on controlled randomized clinical trials.
2. Manufacturing quality improvement and management.
3. Software testing.
In this domains (and in many others) researchers have dire troubles in supporting complex experiments and experiments without fully determined model (fuzzy model, etc.). One of the problems can be stated as how to integrate methods of the statistical data analysis and methods of the data analysis based on formalized knowledge (so called intelligent data analysis).
Following first three major steps in the data analysis are well described and are widely used.
1. Cleaning and organizing the data for analysis (Data Preparation or Data Tiding).
2. Describing the data (Descriptive Statistics).
3. Testing hypotheses (Inferential Statistics).
But for full-blown execution of those steps one needs apriori formal model, which causes specific statistical tests and various constraints. It is a great pity researcher "breaks" the data, that one does not need as input for statistical instruments in the current experiment, but one has those data as part of protocol. In practice with using a modern set of advanced experimental tools one has so much data that we need to break more than 90% of the data from the protocol and more than 99,9% of the whole collected data. In turn intelligent data analysis methods (particularly, ontology-controlled methods) are able to work with almost all raw data. In the seminar we will discuss modern methods of data tiding, information retrieval and knowledge discovery as well as various applications of those methods.
Also the seminar will focus on:
1. Controlling errors in statistical experimenting.
2. Analyzing raw data in form of collections of unstructured texts.
3. Using software tools for integrated automation of experimentation processes.
4. Using software tools for intelligent data analysis.
Everyone interested is invited!
Working language is English.
Videoconference with St. Petersburg office of the LCSR (room 303, 55-2, Sedova st.) will be provided. Guests are free to enter the building.
Guests from Moscow are free to enter to the 5th floor from 17:45 till 18:15. For questions: +7-495-772-95-90 extension number 12244.
Also after the seminar a videotape will be put up after the summary of the presentation.
Date
10 September
18:00
Address
Москва, Мясницкая 22, ауд. 511; Санкт-Петербург, ул. Седова 55-2, ауд. 303