Background: Therapists use a variety of methods to teach skills to children with Autism Spectrum Disorder. Many of these methods require extensive reward protocols. Cold probe testing is a method by which therapists determine the level a child has mastered a skill without any incentives. At the Language and Learning Clinic of a metropolitan area autism center, the first 15 minutes of therapy are spent assessing the status of every skill that is in treatment using this method. The therapist uses a Cold Probe Sheet to ascertain if a skill is mastered. If a skill is mastered then it is moved into maintenance mode but if it is not then it stays in treatment. Information from this sheet is manually collected and later entered into the computer. The data is then analyzed using a standard spreadsheet program. The current process has a number of limitations. First, it is time consuming for therapists to transfer the data from paper to excel spreadsheet. Second, errors can occur when transferring data from manual to digital format. Finally, the clinical supervisor can not access the patients’ results until all of the therapists have input data.
Objectives: The goals of this study were to 1) develop an automated method to process the information collected from the cold probe sheet and 2) automatically transfer the data into spreadsheet.
Methods: We used a human computer interaction perspective to understand the users’needs, this included interviews with therapist and their supervisor. The stake holders were also involved in design cycle. Given their needs we decided to use a digital pen to collect data entered into the cold probe sheet using handwriting recognition technology. Prototype and evaluation cycles were carried out simultaneously. When the system was first deployed, we studied two different groups; one used the new system; the other used the standard manual routine.
Results: The initial handwriting solution had some problems, and the system currently runs a modified version of HWR. This was designed based on semantic context. Our system has a number of strengths. First it maintains treatment practice. The only change is that the therapists must “dock” the pen at least once a day. The new system is quick and automatic. The spreadsheet data easily integrates with other software used in the center (e.g., statistics package). It also provides additional information which may be relevant, such as the time stamp of data input. Initial qualitative data indicates that this system saves hours of time in everyday practice. Another interesting finding is that the system helped to regulate the behavior of the therapists. Therapists are now aware that the system needs discrete symbols and as a result they use better “penmanship” when completing the cold probe sheet.
Conclusions: Overall, the system seems to have helped the therapists in the autism center to reduce their work load and mistake ratio during data entry. In addition the visualization function helps teachers to understand the data they collect. Based on the current system, we will introduce some data mining function in the future.