using behavioral patterns in treating autistic

14
Sudarsun. S, Member, IEEE Varun Kant Vashishtha Avijit Nayak

Post on 19-Oct-2014

3.432 views

Category:

Health & Medicine


1 download

DESCRIPTION

The development of behavioral therapy regimens for autistic patients is relatively challenging as these patients may not be able to express feedback to the applied treatment. The response to a treatment course is mostly estimated qualitatively and with little systematic feedback between therapy and response. Collecting and analyzing data about a patient’s daily activities could yield patterns linking these activities, thereby providing therapists with some foreknowledge of likely possible behavioral outcomes related to their therapies. We propose a method for anomaly detection system, which can monitor behavior patterns of the patient based on the data collected on a daily basis. The knowledge gathered about the patient could prove suggestive of the patient’s feedback to the applied therapy. Upon mining the behavioral patterns, the system could predict the response of a patient to a stimulus, given a list of recently displayed behaviors and/or completed activities. The knowledge thus gathered could also be used to treat other patients of similar disability.

TRANSCRIPT

Page 1: Using Behavioral Patterns In Treating Autistic

Sudarsun. S, Member, IEEE

Varun Kant VashishthaAvijit Nayak

Page 2: Using Behavioral Patterns In Treating Autistic

Brain development disorder Generally discovered during infancy or childhood Impairments in social interaction and communication Restricted and repetitive behavior Self-injury and aggression No specific cure available

Page 3: Using Behavioral Patterns In Treating Autistic

Use of Power Point slides with video segments to teach socio-dramatic play. Train autistic people to point to pictures to aid communication. Simulations use to teach verbal communications or social robots for diagnosis and treatment.

Page 4: Using Behavioral Patterns In Treating Autistic

Substantive variations among cases Patients socially and intellectually not matured No immediate feedback available Low levels of program adherence

Page 5: Using Behavioral Patterns In Treating Autistic

To improve the life of autistic patients by helping the therapist understand more systematically & more immediately what is the effect of therapy. To predict on the fly when a patient will behave appropriately or not.

Page 6: Using Behavioral Patterns In Treating Autistic

Collection of patient’s observable behavior Classification of observed behaviors Using the sequence of observed behavior to determine the underlying mental state. If we have a sequence of behavioral pattern (say A=>B=>C=>D) what is the next most probable activity that patient will perform. This will help in preventing any unwanted situation beforehand.

Page 7: Using Behavioral Patterns In Treating Autistic

Problems we wish to address:1. To estimate the effectiveness of the

treatment from the observations on the patient.

2. Given a pattern of behavioral symptoms, how to predict the upcoming changes in the patient’s behavior.

3. How to detect anomalies in the patient’s behavior.

Page 8: Using Behavioral Patterns In Treating Autistic

Estimating the effectiveness of the therapy:Three sequences:1. The medication event sequence.2. Patient’s feedback event sequence.3. Recorded patient’s observed behavior sequence.Medication event sequence is a controller input that is to be manipulated by patient’s feedback sequence. Patient’s feedback sequence it is not readily available (Hidden). But we have the recorded observations on the patient’s behavior which are dependent on the hidden patient feedback sequence.

Page 9: Using Behavioral Patterns In Treating Autistic

Visualizing the problem as a set of hidden and observed events. Hidden Markov Models (HMM) can be a solution. E.g. deducing weather from a piece of seaweed. If it is “soggy” means wet weatherIf it is “dry” means sunny weatherIf it is “intermediate” – can’t say (probably sunny or rainy) Second useful clue can be the known condition of weather on previous day plus present (observed) seaweed state provide a better forecast.

Page 10: Using Behavioral Patterns In Treating Autistic

Three canonical problems associated with HMM:

1. Given the parameters of the model, how can we compute the probability of a particular output sequence?

2. Given the parameters of the model, how can we determine the most likely sequence of hidden states that could have generated a given output sequence?

3. Given an output sequence or a set of such sequences, find the most likely set of state transition and output probabilities.

Problem 2 is more related to our case. Viterbi algorithm

can be used to solve this problem.

Page 11: Using Behavioral Patterns In Treating Autistic

Prediction of upcoming behavioral change: Any activity dependent on previous

activity/activities. Use HMM to predict the next hidden mental

state transition (e.g. Observed A=>B=>C what is

the probability of next activity to be D, i.e. P(D/A,B,C)).

First order HMM may not be appropriate (we have to allow the next state to be dependant on more than one previous state).

Use of Conditional random field (CRF) as an alternative to nth order HMM.

Page 12: Using Behavioral Patterns In Treating Autistic

Anomaly detection on behavioral patterns: Simplest way is to manually classify the set of normal and abnormal behaviors. Expert classify the behaviors into set of normal and abnormal behaviors. Fingerprinting is a viable technique for the same.Fingerprinting techniques involve two basic operations: • Alignment of the sequence data, where the sequences are aligned to maximum overlap.• Computation of similarity scores.

Page 13: Using Behavioral Patterns In Treating Autistic

Data mining tools may not be self sufficient Effectiveness depends heavily on how well we collect behavioral data Can help in revealing patterns and relationships but not the significance of these patterns This is a theoretical paper so the accuracy can not be analyzed due to lack of experimental data Validation of these concepts with experimental data is a logical extension of this paper.

Page 14: Using Behavioral Patterns In Treating Autistic