Personalized, Model- Based, Functional Acoustic Hearing Diagnostic

Audiometric methods for the detection of hearing loss are often based on the comparison of measurements with standard curves representing the statistical mean of “normal” hearing. These methods suffer from large inter-individual variances in middle-ear transmission and from the individually very different collaboration of the patients, in particular of small children, seriously ill or elderly patients. Thus the specificity of current diagnostic tests on a particular clinical picture as well as the quantitative evaluation is strongly restricted. This holds true especially for differential diagnosis between inner-ear and neural impairments since the middle-ear comes first in the measurement chain. The author suggests a model-based approach based on objective measurements, e.g. wide-band tympanometry. Several numerical middle-ear models are known in hearing research, but by now do not contribute to individual, patient-specific diagnostics. Their use could increase the specificity of routine diagnostic tests on common middle-ear diseases and would allow an objective quantification of the individual pathologies. 

Research Idea and Context

Starting Point:

Audiometric methods for the detection of hearing loss are often based on the comparison of measurements with standard curves representing the statistical mean of “normal” hearing. Because of the large inter-individual variances in middle-ear transmission, both the specificity of current diagnostic tests on a particular clinical picture as well as the quantitative evaluation are strongly restricted, so that often only a dichotomous yes-or-no decision with low specificity is possible. This holds true especially for differential diagnosis between inner-ear and neural impairments since the middle-ear comes first in the measurement chain. Today, several numerical middle-ear models are known in hearing research, but by now do not contribute to individual, patient-specific, quantitative diagnostics. 

Open Question:

Has the time come to take the plunge into a quantitative and personalized diagnosis based on objective, non-invasive measurement data and individual model-measurement fitting?

Relevance:

Objective measurement methods provide more reliable data than subjective audiometry tests because they are independent from the individually very different collaboration of the patients, in particular, of small children, seriously ill or elderly patients. Based on those objective measurements (e.g. wide band tympanometry, etc.) a model-based evaluation significantly increases the specificity of routine diagnostic tests on common middle-ear diseases and allows an objective quantification of the individual pathologies. Thus, positive or negative trends are detected earlier in the course of the illness and the therapy decision is less dependent on the subjective, medical judgement than on the objective, quantitative findings and therefore promises to be specifically tailored and personalized in the sense of patient well-being. 

In relation to newborn hearing screening, a model-based determination of individual middle-ear transmission enables to eliminate the middle-ear influence on OAE measurements. As a result, the current 15% false-positive screening results would decrease to approximately 2-5% according to own estimations. This would avoid unnecessary stress in children and their parents, it would reduce the number of follow-up examinations and exploit a considerable saving potential in the health care system.  

The model-based approach also opens up the possibility for a non-invasive measurement of the intracranial pressure since the middle-ear transmission shows a systematic dependence on the perilympathic/intracranial pressure and thus can be utilized as a non-invasive sensor for the intracranial pressure. That would be a great innovation in clinical practice with great ethical benefits, since to date, intracranial pressure has to be measured exclusively invasive via a complication-prone surgical procedure, in which a pressure probe, generally without image-giving CT feedback, has to be maneuvered through healthy brain tissue into the ventricular system of the brain. However, a closed-meshed or continuous monitoring is essential for many neurological and neurosurgical syndromes to avoid the feared complication of "herniation” which often causes death.

Objectives

Major Gains

The model-based diagnosis is a novel approach in otological diagnostics. Being explored here for the first time, it presents a true chance to change clinical routine diagnostics to a quantitative, objective and personalized diagnosis. This may serve as a prototype for functional diagnostic developments in other fields.

Potential Bottlenecks

One of the main challenges is the clustering of the parameters. On the one hand, the order of the fitting problem has to be reduced, so that a reliable determination is possible. On the other hand, the individual parameters have to be sufficiently specific to notify local anatomical pathologies. It also remains challenging to which degree of detail the exclusively non-invasive measurement data can be exploited for the estimation of the hidden anatomy-specific parameters with the aid of fuzzy arithmetic to achieve a valid and robust individual middle-ear model.