In response to the guidelines of the PAR-18-352 Funding Opportunity Announcement, this project is proposing a data and analysis method for the reliable and accurate personalized characterization of tinnitus. Bothersome tinnitus is an enormous source of suffering and disability. It is estimated that nearly 15% of the general public ? over 50 million Americans ? experience some form of tinnitus. The proposed study will assess a new methodology to combine ecological momentary assessment (EMA)-based, intensely longitudinal data with new analytical techniques. The study hypothesis is that the capture of tinnitus patients? individual profiles will allow prediction of the currently puzzling heterogeneity of both treatment response and neuroimaging findings. This will be a single arm, longitudinal study design with repeat assessments, and 110 participants will be recruited from Washington University. Participants will be expected to complete 4 EMA surveys via smartphone every day for 3 weeks for a total of 84 pre-CBT surveys. Participants will also undergo resting-state functional connectivity MRI (rs-fcMRI) prior to the start of CBT. Tinnitus offers a particularly appropriate condition to illustrate the utility of improved and reliable personalized assessment because of the heterogeneity in both neuroimaging and treatment findings.
The Specific Aims are: (1) To obtain person-specific drivers of tinnitus through personalized ambulatory assessment and ML- DSEM analyses. (2) To examine the relationship between ML-DSEM-defined drivers of tinnitus and patient response to cognitive behavioral therapy (CBT). (3) To examine the association between ML-DSEM-defined drivers of tinnitus, treatment response, and neuroimaging. The successful conduct of the proposed research will advance tinnitus research through improved patient assessment and data analytical techniques, which will result in greater efficacy of clinical trials and move patient care toward personalized medicine. Such an approach is widely applicable to health conditions that are heterogeneous and widely variable over time. Eventually, a clinician could order the acquisition of longitudinal data through a website or app, which would then render an informative personalized model based on the new analytical techniques described in this research. In this way, the end result of this research could be the true realization of personalized medicine in clinical practice.
Smartphones allow for the collection of real-time symptoms from patients in their natural setting (e.g., ?How do you feel right now??) repeatedly over time and can be more accurate than assessments that ask patients to describe their symptoms over a specific time period (e.g., ?In the last two weeks, how have you felt?). Using new and advanced statistical techniques, researchers can test the smartphone-captured data to see which variables might cause others over time for each individual. This capture of real-time symptoms combined with new statistical techniques can define personalized profiles, which may help predict individual response to treatment, and contribute to the development of personalized medicine.