Risk Assessment for Early Childhood Caries ECC, RAE - PowerPoint PPT Presentation

1 / 14
About This Presentation
Title:

Risk Assessment for Early Childhood Caries ECC, RAE

Description:

What is Knowledge Discovery and Data Mining (KDD) ... tape unreadable, no raw exam data. no S-ECC, but other indices (deft, dfs, mi-deft, mi-dfs, etc) ... – PowerPoint PPT presentation

Number of Views:130
Avg rating:3.0/5.0
Slides: 15
Provided by: janewei
Category:

less

Transcript and Presenter's Notes

Title: Risk Assessment for Early Childhood Caries ECC, RAE


1
Risk Assessment for Early Childhood Caries (ECC),
RAE
  • Support DHHS/NIH/NIDCR,NCMHHD
  • U54 DE 14251

2
Aims
  • Knowledge Discovery Data Mining (KDD) to
    predict Early Childhood Caries (ECC) with
    California Oral Health Needs Assessment (COHNA)
  • Test model in Health Kids,
    Healthy Teeth (HKHT)
  • Test models in FV RCT (Proj 1)

3
What is Knowledge Discovery and Data Mining (KDD)?
  • Semi-automatic discovery of patterns,
    associations, anomalies, and statistically
    significant structures in data MIT Tech Review
    (2001)
  • Interface of
  • Artificial Intelligence Machine Language
  • Computer Science Statistics
  • Association for Computing Machinery Special
    Interest Group on Knowledge Discovery in Data and
    Data Mining (ACM SIGKDD sponsors KDD Cup)

4
KDD Learning Tools
  • Supervised
  • Regression
  • Classification
  • k nearest neighbor
  • Trees (CART, MART, boosting, bagging)
  • Neural Networks
  • Generalized Additive Models
  • Multivariate Adaptive Regression Splines (MARS)
  • Support Vector Machines
  • Random Forests
  • Unsupervised
  • Hierarchical clustering
  • K-means

5
Modeling Methods
Logistic Regression
Neural Networks
Decision Trees
6
Validation
  • Split sample (70 training / 30 validation)
  • Validation estimates unbiased misclassification
  • K-fold Cross Validation
  • Mean squared error

7
Data Quality
8
Critique
  • Q N (2000) too small for data mining?
  • A 2001 KDD Cup
  • N1900 (49 events) 130,000 variables
  • Q Whats the practical application?
  • A User friendly risk assessment tools
  • Q Variable data quality, missingness?
  • A

9
(No Transcript)
10
Results
  • ECC health disparities differ by race/ethnicity x
    Head Start status interaction (Shiboski, Gansky
    et al JPHD 2003)
  • KDD tools for dental research
  • (Gansky Adv Dent Res 2003?)
  • 16 / 283 ZIP codes changed 1993 gt 1998

11
Modified Plans
  • COHNA
  • tape unreadable, no raw exam data
  • no S-ECC, but other indices
  • (deft, dfs, mi-deft, mi-dfs, etc).
  • HKHT
  • CMS (HCFA) funded 1 year pilot (100K)
  • some funds procured, more funding sought
  • Arizona Study
  • SYHC IOHP Data

12
Questions
  • Should CANDO Year 3 funds (100K) be used for
    HKHT child dental exams?

13
DISCUSSION
14
(No Transcript)
Write a Comment
User Comments (0)
About PowerShow.com