assessing the readability of electronic health records lan voba si 561 natural language processing
TRANSCRIPT
ASSESSING THE READABILITY OF ELECTRONIC HEALTH RECORDSLan VoBa
SI 561 Natural Language Processing
Overview of EHRs• What is an EHR?• How are they being used?• What are the benefits?• What are the challenges?
Motivation• How “patient-friendly” are doctors’ notes, letters, etc.?• How difficult to read and understand those documents?• Related work
• “Applying Multiple Methods to Assess the Readability of a Large Corpus of Medical Documents” (Wu et al)
• Readability formulas used to assess other health-related texts• Automated Readability Index• New Dale-Chall Formula
The Data• UMHS
• 120 clinic locations and offices• 45,000 inpatient hospital stays• 1.8 million outpatient visits and surgeries
• The records• Freetext• Preparation• 4,133 Letters• 18,217 non-letters
• Challenges• Limited access• Poor metadata
Automated Readability Index
Age Group Grade Level
5 – 6 years old Kindergarten
6 – 7 years old First Grade
7 – 8 years old Second Grade
8 – 9 years old Third Grade
9 – 10 years old Fourth Grade
10 – 11 years old Fifth Grade
11 – 12 years old Sixth Grade
Age Group Grade Level
12 – 13 years old Seventh Grade
13 – 14 years old Eighth Grade
14 – 15 years old Ninth Grade
15 – 16 years old Tenth Grade
16 – 17 years old Eleventh Grade
17 – 18 years old Twelfth Grade
18 – 22 years old College
New Dale-Chall Formula
Score Meaning
4.9 and Below Grade 4 and Below
5.0 to 5.9 Grades 5 – 6
6.0 to 6.9 Grades 7 – 8
7.0 to 7.9 Grades 9 – 10
8.0 to 8.9 Grades 11 – 12
9.0 to 9.9 Grades 13 – 15 (College)
10 and Above Grades 16 and Above (College Graduate)
Method
Metadata
LettersAutomated Readability
Index
New Dale-Chall
Formula
Non-Letters
Scores
Results
Future Work• Wu et al• More documents• Remove “headers”• Compare scores to human evaluators to confirm
assessment
Thank you!Questions? Comments?