1 Virtual Project Members Matthew Harris Project Leader

1

Virtual Project Members • Matthew Harris, Project Leader – Senior, ECE • Allison Huber – 2 nd year Au. D. graduate student • Matthew Hunold – Senior, ECE • Allison Witte – 2 nd year Au. D. graduate student • Aaron Whiteman – Freshman, First Year Engineering 2

Project Partner • Dr. Joshua Alexander, Ph. D. , CCC-A Assistant Professor- Dept. of Speech, Language, & Hearing Sciences • SLAC Team sponsored by Motorola http: //acsspirit. com/walkietalkieimages/jpg/Motorola_blue_Horz. jpg 3 the MOTOROLA and the Stylized M Logo are registered in the US Patent & Trademark Office. All other product or service names are property of their respective owners. ©Motorola, Inc. 2006.

Problem Identification • The Doctor of Audiology (Au. D) students at Purdue require a more practical and realistic way of practicing masking skills. – These skills are an important part of the clinical test protocol, and are used to ensure accuracy of results. • Limited availability of subjects to apply clinical skills 4

Project Overview • Project Goal: create a software simulation of an audiological examination to help students master clinical audiology masking skills • Users perform hearing tests on simulated patients to identify a certain pathology requiring masking. 5

Overview: Sound Transmission Brandyoumedialab. net 6

Audiological Assessment Typical Protocol: • Threshold, or the lowest level a patient can hear 50% of the time, is determined via: – Air Conduction (AC) • through earphones – Bone Conduction (BC) • by an oscillator that vibrates the skull www. audiology-iicm. net, sensograph. com, fortunecity. com 7

• If there is a large difference in hearing between the two ears, or between AC and BC of one ear: masking may be needed www. hearcom. eu 8

What is masking? • Uses narrowband noise to occupy the better hearing (non-test) ear – Prevents loud tones from crossing over to the other ear and contributing. • Enables attainment of true threshold of the poorer ear, without the better ear’s “help” 9

Specification Development - User • The software user should be able to: – Correctly acquire unmasked AC and BC thresholds and make appropriate clinical masking decisions – Integrate case history information and obtained audiometric results to diagnose different hearingrelated pathologies – Visualize effective masking levels through the plateau plot 10

Specification Development - Product • In order to meet the user goals, the software should: – Display initial patient information (case history) – Display a computer generated patient – Have an audiogram template on which thresholds can be plotted – Simulate masking for the virtual patient, and show corresponding plateau plot – Record student made audiogram for grading • Each bullet will correspond to a subdivision of the main window 11

Software overview • Contains 2 modes: Practice and Evaluation – Practice: student selects patient by expected (or random) pathology, and matches obtained audiogram with expected results – Evaluation: student receives unknown patient pathology • Student gets to perform same test protocol used in clinic situations 12

Software layout & Plateau Plot 13

Previous work • Fall 2010 project team delivered version 1. 0 – Bare bones practice mode – Incomplete evaluation mode – Includes visual plateau plot • Work continued in Spring 2011 – Subversion capabilities – Four new patient images 14

Conceptual Design: Randomizing Pathologies • Previous versions only have one possible outcome per pathology • We want a more realistic simulation • What exactly are we randomizing? • How do we randomize it? – How far? What range? Which model? – Cycle through five predetermined variations? • How do we put it into code? 15

Implementing Randomness Criteria Base Case (One Pathology) Several Pathologies Randomization Algorithm Educational: Can students learn from it? - + Accurate: Would the same thing happen in nature? - Difficulty: Are we able to model and code our variations? + - Repeatable: Will the student be able to pick up on patterns? + - Overall: Which would make the most positive contribution to the project? - + 16

Math Algorithm Approach • We chose to develop models for each pathology and apply calculated randomness • A good starting point was to correctly model “normal hearing”. Two benefits: – Could easily make half the audiogram if the pathology only affects one ear – All conditions depend on certain aspects of a patient’s normal hearing 17

Fall 2011 Summary of Semester Progress • The clinicians analyzed the 10 pathologies and created range descriptions for each – Presbycusis, Unilateral and Bilateral Otosclerosis, Unilateral and Bilateral Otitis Media, Meniere’s Disease, Acoustic Tumor, Noise Induced Trauma, Microtia, Ossicular Discontinuity. • Programmers determined algorithms based on these descriptions 18

Semester Progress cont’d • An HTML document was created to visually display the student answers for instructor review • A workshop was set up to gain feedback from undergraduate and Au. D students – Masking verification – Evaluation mode output accuracy – Bug checking 19

Semester Progress cont’d 20

Evaluation Report layout • Thresholds laid out in a table form • “Theoretical” Thresholds generated by the program are displayed in white • Student answers are color coded – Green : Correct threshold and test settings – [ Red ] : Audiometer settings incorrect – ( Yellow ) : Incorrect threshold acquired • Test settings appear in a popup box once the instructor clicks on a certain student answer 21

Randomizing Normal Hearing • The randomized normal hearing pathology is the heart of all of the other pathologies • The assumption is made that anyone with a hearing disorder at one time had, or could have had, normal hearing • Each pathology situation shapes the normal hearing pathology into a form specific to the hearing disorder 22

Randomizing Normal Hearing • Conditions that determine Normal Hearing: – All values must fall in between the range of -10 and 25 d. B – AC (Air Conduction) between two neighboring frequencies must be within 10 d. B – BC (Bone Conduction) between two neighboring frequencies must be within 10 d. B – AC must be within 10 d. B of BC for the same ear and frequency – BC must not exceed AC for the same ear and frequency – BC between both ears for the same frequency must be within + or – 5 d. B (IA approx. 0) – AC for both ears at the same frequency must be within 10 d. B 23

Evaluation Mode Report (Why HTML? ) • Pros – Simple to output text and still provide a graphical display – Do not need another program to display values, a web browser will do the job • Cons – No encryption. A savvy user can easily hack their report and change values prior to submission to the instructor 24

Plotting Thresholds in Eval Mode • Audiometer settings at the time of plotting determine accuracy of student response – Transducer – Frequency – Interrupt – Intensity (d. B) – Left/Right ear 25

HTML Value Swapping • The template HTML file has raw entries that need to be replaced upon saving results • The template HTML file is searched for particular keywords that correspond to tags (IDs) that identify the location for the appropriate data • The keywords are replaced by the student and program generated values to provide a complete evaluation report 26

Workshop 12/1/2011 • Survey – Does the program seem intuitive? Is it easy to use/navigate? – Were there any bugs or problems in the program during your session? Was any information on the screen incorrect? – Did the patient/case you used have a fitting HL for the given disorder? Were all of the values correct, including masking values? – Any other comments or areas for future improvement?

Workshop 12/1/2011 • Feedback – Undergrads: Need simpler audiometer layout – Grad Students: Easy audiometer, same as clinic – System crashed when clicking buttons too quickly – Pathologies appeared to match audiograms – Hand raise – Need bigger audiogram/hard to plot – Masking calculator/notes

Semester Timeline • Weeks 1 -2 – Get new members familiar with the project and semester goals • Weeks 3 -7 – Design and implement randomization process • Weeks 8 -11 – Evaluation Mode revamp • Weeks 12 -14 – Some testing of Randomization and Evaluation Mode • Week 15 – Design Review 29

DEMO! 30

Transition Report • • • • • Recommended work for Spring 2012 Semester Eval Report Output: Need to provide student name, assignment, , and course. Add ID’s to the HTML tags that contain the text and replace the strings Need to provide accurate percentages for answers on output HTML Need to fix masking values always showing up wrong Need to place Frequency on the X axis instead of Y axis (to match an audiogram) Need to provide space to put additional information as to why an answer was wrong Need to ensure popup boxes are providing the correct information about audiometer settings Need to provide a printable version of the eval report Need to query the student as to what they believe the pathology is upon saving results Patient: The patient raises the wrong hand (A tone presented to the right ear should raise the patients right hand, not the right hand as seen by clinician) Images are needed on the patient window to show which transducers are being used on the patient Audiogram: The audiogram window needs to be bigger height-wise, the standard is 20 d. B has the same distance as an octave. Crashes: Pressing the present button rapidly may cause the program to crash. 31

Questions? 32