MEDICAL ERROR REPORTING AND ANALYSIS

1
MEDICAL ERROR REPORTING AND ANALYSIS

• Vijaya Gotla
• UmaDevi Bandaru
• Lavanya Gundamaraju

2
MERA

• A Decision support system
• Two major functions
• Error Report- The user reports an error by
  entering its details.
• Error Analysis- User requests the information and
  statistics regarding a particular error.
• Basically a research than an application
• Incorporates
• Distributed agent technology
• Data mining and pattern matching
• Machine learning

3
Software Agents

• Inhabits some complex dynamic environment
• Senses and acts autonomously in this environment
• Realize a set of goals or tasks for which they
  are designed

4
Agent Characteristics

• Autonomous
• Intelligent
• Adaptive/learning
• Capable of reasoning
• Mobile
• Persistent
• Goal oriented
• Communicative/collaborative
• Flexible
• Open

5
Classification Of Agents

• Based on nature of task performed
• -User Agent, Service Agent
• Based on nature source of intelligence
• -Rule-based using user programmed rules
• -Knowledge base programmed by knowledge engineer
• -Machine Learning that program themselves
• Based on location/mobility
• -Fixed Agents, Mobile agents

6
Learning Agents

• Can acquire and apply knowledge problem-solving
  methods
• can learn, adjust that knowledge based on
  interaction with its environment
• learns rules for classifying and archiving data

7
Machine learning

• Machine learning makes data mining easier in case
  of large amounts of data
• Learn a computational structure
• Functions, problem solving systems, grammars and
  finite state machines.
• LearningSupervised Unsupervised

8
ALGORITHMS

• Predictive Methods
• Classification
• Regression
• Deviation Detection
• Descriptive Methods
• Clustering
• Association Rule Discovery
• Sequential Pattern Discovery

9
Projects in Agent Technology

• Interface agents for the electronic-mail domain
• Monitors the actions of the user
• Uses Memory based Reasoning algorithm.
• Compares and matches
• Predicts an action

10
Projects(Cont.)

• Personalized Information Filtering
• Build a set of adaptive autonomous interface
  agents that inhabit the user's computer
• Uses the Feedback and Genetic algorithms
• Options on Feedback
• Creates new profile
• Directs to existing profile

11
Inference

• Agents are autonomous programmed components
• Agents can be used to monitor user actions
• User can exploit the learning ability of the
  agent to ease his job
• Agents can collaborate with each other

12
What is in MERA

• Distributed, Component based, Agent technology
• Learning Agent
• Association Based Algorithm
• Collaboration in Agents
• Not Peer Agents

13
System Framework

Service Agent
DB
User Agent
14
Drawbacks

• Internal state of the agent is abstract to the
  user
• Application specific Agent development
• Agent Reusability
• Slow Learning curve
• Require sufficient amount of time before they can
  be of any use
• Greatest drawbacks to machine learning is that
  knowledge acquired while learning one skill does
  not enhance the learning of other related skills

15
Thank You.