Universal automated classification of the acoustic startle reflex using machine learning.

The startle reflex (SR), a robust, motor response elicited by an intense auditory, visual, or somatosensory stimulus has been widely used as a tool to assess psychophysiology in humans and animals for almost a century in diverse fields such as schizophrenia, bipolar disorder, hearing loss, and tinnitus. Previously, SR waveforms have been ignored, or assessed with basic statistical techniques and/or simple template matching paradigms. This has led to considerable variability in SR studies from different laboratories, and species. In an effort to standardize SR assessment methods, we developed a machine learning algorithm and workflow to automatically classify SR waveforms in virtually any animal model including mice, rats, guinea pigs, and gerbils obtained with various paradigms and modalities from several laboratories. The universal features common to SR waveforms of various species and paradigms are examined and discussed in the context of each animal model. The procedure describes common results using the SR across species and how to fully implement the open-source R implementation. Since SR is widely used to investigate toxicological or pharmaceutical efficacy, a detailed and universal SR waveform classification protocol should be developed to aid in standardizing SR assessment procedures across different laboratories and species. This machine learning-based method will improve data reliability and translatability between labs that use the startle reflex paradigm.

Development of ebselen for the treatment of sensorineural hearing loss and tinnitus.

The global impact of hearing loss and related auditory dysfunction including tinnitus and hyperacusis on human health is significant and growing. A substantial body of literature has found that these hearing diseases and disorders result from significant number of genetic variations and molecular mechanisms. Investigational new drugs have been tested and several approved drugs have been repurposed in clinical trials, but no therapeutics for any auditory related indication have been FDA approved. A unique investigational new drug called ebselen (SPI-1005), that is anti-inflammatory and neuroprotective, has been shown to reduce noise-induced and aminoglycoside-induced hearing loss in animals. Multiple phase 2 clinical trials have demonstrated the safety and efficacy of SPI-1005 treatment in Meniere's disease and acute noise-induced hearing loss. SPI-1005 is currently being tested to prevent and treat tobramycin-induced ototoxicity in cystic fibrosis patients with acute lung infections. This review summarizes the published and presented data involving SPI-1005 and other drugs being tested to prevent or treat sensorineural hearing loss. Additionally, recent clinical data showing the relationship between pure tone audiometry and words-in-noise test results in a Meniere's disease are presented, which may have larger implications for the field of hearing research.

Development of Tinnitus and Hyperacusis in a Mouse Model of Tobramycin Cochleotoxicity.

Aminoglycosides (AG) antibiotics are a common treatment for recurrent infections in cystic fibrosis (CF) patients. AGs are highly ototoxic, resulting in a range of auditory dysfunctions. It was recently shown that the acoustic startle reflex (ASR) can assess behavioral evidence of hyperacusis and tinnitus in an amikacin cochleotoxicity mouse model. The goal of this study was to establish if tobramycin treatment led to similar changes in ASR behavior and to establish whether ebselen can prevent the development of these maladaptive neuroplastic symptoms. CBA/Ca mice were divided into three groups: Group 1 served as a control and did not receive tobramycin or ebselen, Group 2 received tobramycin (200 mg/kg/s.c.) and the vehicle (DMSO/saline/i.p.) daily for 14 continuous days, and Group 3 received the same dose/schedule of tobramycin as Group 2 and ebselen at (20 mg/kg/i.p.). Auditory brainstem response (ABR) and ASR hearing assessments were collected at baseline and 2, 6, 10, 14, and 18 weeks from the start of treatment. ASR tests included input/output (I/O) functions which assess general hearing and hyperacusis, and Gap-induced prepulse inhibition of the acoustic startle (GPIAS) to assess tinnitus. At 18 weeks, histologic analysis showed predominantly normal appearing hair cells and spiral ganglion neuron (SGN) synapses. Following 14 days of tobramycin injections, 16 kHz thresholds increased from baseline and fluctuated over the 18-week recovery period. I/O functions revealed exaggerated startle response magnitudes in 50% of mice over the same period. Gap detection deficits, representing behavioral evidence of tinnitus, were observed in a smaller subset (36%) of animals. Interestingly, increases in ABR wave III/wave I amplitude ratios were observed. These tobramycin data corroborate previous findings that AGs can result in hearing dysfunctions. We show that a 14-day course of tobramycin treatment can cause similar levels of hearing loss and tinnitus, when compared to a 14-day course of amikacin, but less hyperacusis. Evidence suggests that tinnitus and hyperacusis might be common side effects of AG antibiotics.

Machine learning, waveform preprocessing and feature extraction methods for classification of acoustic startle waveforms.

The acoustic startle response (ASR) is an involuntary muscle reflex that occurs in response to a transient loud sound and is a highly-utilized method of assessing hearing status in animal models. Currently, a high level of variability exists in the recording and interpretation of ASRs due to the lack of standardization for collecting and analyzing these measures. An ensembled machine learning model was trained to predict whether an ASR waveform is a startle or non-startle using highly-predictive features extracted from normalized ASR waveforms collected from young adult CBA/CaJ mice. Features were extracted from the normalized waveform as well as the power spectral density estimates and continuous wavelet transforms of the normalized waveform. Machine learning models utilizing methods from different families of algorithms were individually trained and then ensembled together, resulting in an extremely robust model.•ASR waveforms were normalized using the mean and standard deviation computed before the startle elicitor was presented•9 machine learning algorithms from 4 different families of algorithms were individually trained using features extracted from the normalized ASR waveforms•Trained machine learning models were ensembled to produce an extremely robust classifier.

Ebselen attenuates tobramycin-induced ototoxicity in mice.

BACKGROUND: Cystic fibrosis patients are often adminstered tobramycin to treat pulmonary infections. Unfortunately, a common side effect is hearing loss, which can fluctuate. Ebselen has known anti-inflammatory properties and could reduce the incidence and severity of tobramycin-induced hearing loss. METHODS: In vitro: neonatal cochlear cultures were treated with tobramycin or cotreated with tobramycin and ebselen for 3 days. In vivo: adult mice were injected with tobramycin or tobramycin and ebselen for 14 days. ABRs were collected in a repeated measures design until 56 days after treatments. ABR threshold shifts were analyzed and a novel cochleotoxic criteria applied to determine the incidence of ototoxicity. Cochlear immunohistology was analyzed for IHC and OHC loss. RESULTS: Tobramycin leads to significant IHC and OHC loss in cochlear explant cultures. Ebselen co-treatment at 1:20 concentrations resulted in significant otoprotection. Tobramycin leads to significant ABR threshold shifts that are ameliorated by ebselen co-treatment. Hearing loss did not correlate with significant IHC or OHC loss. CONCLUSIONS: This mouse model of tobramycin-induced ototoxicity is clinically relevant in that it results in an incidence and severity of hearing loss recently documented in clinic. The in vitro experiments show that tobramycin kills hair cells and that ebselen co-treatment can attenuate this ototoxicity. The in vivo model shows tobramycin-induced hearing loss is ameliorated by ebselen co-treatment, but this is not explained by concomitant hair cell loss. These preclinical data support the testing of ebselen in CF patients receiving tobramycin treatment.

A Novel Mouse Model of Aminoglycoside-Induced Hyperacusis and Tinnitus.

Aminoglycosides (AG) such as amikacin are commonly used in cystic fibrosis patients with opportunistic pulmonary infections including multi-drug resistant mycobacterium tuberculous and non-tuberculous mycobacterium. Unfortunately, this class of drugs is known to cause peripheral damage to the cochlea leading to hearing loss that can fluctuate and become permanent over time or multiple exposures. However, whether amikacin can lead to central auditory dysfunction like hyperacusis (increased sensitivity to sound) or tinnitus (perception of sound in the absence of acoustic stimulation) is not well-described in the literature. Thus, an animal model needs to be developed that documents these side effects in order to develop therapeutic solutions to reduce AG-induced auditory dysfunction. Here we present pioneer work in mice which demonstrates that amikacin can lead to fluctuating behavioral evidence of hyperacusis and tinnitus as assessed by the acoustic startle reflex. Additionally, electrophysiological assessments of hearing via auditory brainstem response demonstrate increased central activity in the auditory brainstem. These data together suggest that peripheral AG-induced dysfunction can lead to central hyperactivity and possible behavioral manifestations of hyperacusis and tinnitus. Importantly, we demonstrate that ebselen, a novel investigational drug that acts as both an antioxidant and anti-inflammatory, can mitigate AG-induced hyperacusis.

Automated classification of acoustic startle reflex waveforms in young CBA/CaJ mice using machine learning.

BACKGROUND: The acoustic startle response (ASR) is a simple reflex that results in a whole body motor response after animals hear a brief loud sound and is used as a multisensory tool across many disciplines. Unfortunately, a method of how to record, process, and analyze ASRs has yet to be standardized, leading to high variability in the collection, analysis, and interpretation of ASRs within and between laboratories. NEW METHOD: ASR waveforms collected from young adult CBA/CaJ mice were normalized with features extracted from the waveform, the resulting power spectral density estimates, and the continuous wavelet transforms. The features were then partitioned into training and test/validation sets. Machine learning methods from different families of algorithms were used to combine startle-related features into robust predictive models to predict whether an ASR waveform is a startle or non-startle. RESULTS: An ensemble of several machine learning models resulted in an extremely robust model to predict whether an ASR waveform is a startle or non-startle with a mean ROC of 0.9779, training accuracy of 0.9993, and testing accuracy of 0.9301. COMPARISON WITH EXISTING METHODS: ASR waveforms analyzed using the threshold and RMS techniques resulted in over 80% of accepted startles actually being non-startles when manually classified versus 2.2% for the machine learning method, resulting in statistically significant differences in ASR metrics (such as startle amplitude and pre-pulse inhibition) between classification methods. CONCLUSIONS: The machine learning approach presented in this paper can be adapted to nearly any ASR paradigm to accurately process, sort, and classify startle responses.

Addressing variability in the acoustic startle reflex for accurate gap detection assessment.

The acoustic startle reflex (ASR) is subject to substantial variability. This inherent variability consequently shapes the conclusions drawn from gap-induced prepulse inhibition of the acoustic startle reflex (GPIAS) assessments. Recent studies have cast doubt as to the efficacy of this methodology as it pertains to tinnitus assessment, partially, due to variability in and between data sets. The goal of this study was to examine the variance associated with several common data collection variables and data analyses with the aim to improve GPIAS reliability. To study this the GPIAS tests were conducted in adult male and female CBA/CaJ mice. Factors such as inter-trial interval, circadian rhythm, sex differences, and sensory adaptation were each evaluated. We then examined various data analysis factors which influence GPIAS assessment. Gap-induced facilitation, data processing options, and assessments of tinnitus were studied. We found that the startle reflex is highly variable in CBA/CaJ mice, but this can be minimized by certain data collection factors. We also found that careful consideration of temporal fluctuations of the ASR and controlling for facilitation can lead to more accurate GPIAS results. This study provides a guide for reducing variance in the GPIAS methodology - thereby improving the diagnostic power of the test.

Variable Effects of Acoustic Trauma on Behavioral and Neural Correlates of Tinnitus In Individual Animals.

The etiology of tinnitus is known to be diverse in the human population. An appropriate animal model of tinnitus should incorporate this pathological diversity. Previous studies evaluating the effect of acoustic over exposure (AOE) have found that animals typically display increased spontaneous firing rates and bursting activity of auditory neurons, which often has been linked to behavioral evidence of tinnitus. However, only a subset of studies directly associated these neural correlates to individual animals. Furthermore, the vast majority of tinnitus studies were conducted on anesthetized animals. The goal of this study was to test for a possible relationship between tinnitus, hearing loss, hyperactivity and bursting activity in the auditory system of individual unanesthetized animals following AOE. Sixteen mice were unilaterally exposed to 116 dB SPL narrowband noise (centered at 12.5 kHz) for 1 h under ketamine/xylazine anesthesia. Gap-induced prepulse inhibition of the acoustic startle reflex (GPIAS) was used to assess behavioral evidence of tinnitus whereas hearing performance was evaluated by measurements of auditory brainstem response (ABR) thresholds and prepulse inhibition PPI audiometry. Following behavioral assessments, single neuron firing activity was recorded from the inferior colliculus (IC) of four awake animals and compared to recordings from four unexposed controls. We found that AOE increased spontaneous activity in all mice tested, independently of tinnitus behavior or severity of threshold shifts. Bursting activity did not increase in two animals identified as tinnitus positive (T+), but did so in a tinnitus negative (T-) animal with severe hearing loss (SHL). Hyperactivity does not appear to be a reliable biomarker of tinnitus. Our data suggest that multidisciplinary assessments on individual animals following AOE could offer a powerful experimental tool to investigate mechanisms of tinnitus.

An improved approach to separating startle data from noise.

BACKGROUND: The acoustic startle reflex (ASR) is a rapid, involuntary movement to sound, found in many species. The ASR can be modulated by external stimuli and internal state, making it a useful tool in many disciplines. ASR data collection and interpretation varies greatly across laboratories making comparisons a challenge. NEW METHOD: Here we investigate the animal movement associated with a startle in mouse (CBA/CaJ). Movements were simultaneously captured with high-speed video and a piezoelectric startle plate. We also use simple mathematical extrapolations to convert startle data (force) into center of mass displacement ("height"), which incorporates the animal's mass. RESULTS: Startle plate force data revealed a stereotype waveform associated with a startle that contained three distinct peaks. This waveform allowed researchers to separate trials into 'startles' and 'no-startles' (termed 'manual classification). Fleiss' kappa and Krippendorff"s alpha (0.865 for both) indicate very good levels of agreement between researchers. Further work uses this waveform to develop an automated startle classifier. The automated classifier compares favorably with manual classification. A two-way ANOVA reveals no significant difference in the magnitude of the 3 peaks as classified by the manual and automated methods (P1: p=0.526, N1: p=0.488, P2: p=0.529). COMPARISON WITH EXISTING METHOD(S): The ability of the automated classifier was compared with three other commonly used classification methods; the automated classifier far outperformed these methods. CONCLUSIONS: The improvements made allow researchers to automatically separate startle data from noise, and normalize for an individual animal's mass. These steps ease inter-animal and inter-laboratory comparisons of startle data.

Age effects on tinnitus and hearing loss in CBA/CaJ mice following sound exposure.

Tinnitus is a maladaptive neuropathic condition that develops in humans and laboratory animals following auditory insult. In our previous study we demonstrated that sound exposure leads to development of behavioral evidence of tinnitus in a sample of exposed mice. However, this tinnitus mouse model did not account for long-term maladaptive plasticity or aging, factors that are commonly linked to the human tinnitus population. Therefore the same group of mice was monitored for tinnitus for 360 days post exposure. Tinnitus was assessed behaviorally by measuring gap-induced pre-pulse suppression of the acoustic startle (GPIAS). Cochlear histology was performed on both control (unexposed) and experimental mice to determine whether sound exposure caused any evident cochlear damage. We found that 360 days after exposure the vast majority of exposed mice exhibited similar gap detection deficits as detected at 84 days post exposure. These mice did not demonstrate significant loss of inner/outer hair cells or spiral ganglion neurons compared to the control sample. Lastly, we demonstrated that GPIAS deficits observed in exposed animals were unlikely exclusively caused by cochlear damage, but could be a result of central auditory maladaptive plasticity. We conclude that CBA/CaJ mice can be considered a good animal model to study the possible contribution of age effects on tinnitus development following auditory insult.

Development of tinnitus in CBA/CaJ mice following sound exposure.

Tinnitus, the perception of a sound without an external acoustic source, is a complex perceptual phenomenon affecting the quality of life in 17% of the adult population. Despite its ubiquity and morbidity, the pathophysiology of tinnitus is a work in progress, and there is no generally accepted cure or treatment. Development of a reliable common animal model is crucial for tinnitus research and may advance this field. The goal of this study was to develop a tinnitus mouse model. Tinnitus was induced in an experimental group of mice by an exposure to a loud (116 dB sound pressure level (SPL)) narrow band noise (one octave, centered at 16 kHz) during 1 h under anesthesia. The tinnitus was then assessed behaviorally by measuring gap induced suppression of the acoustic startle reflex. We found that a vast majority of the sound-exposed mice (86%) developed behavioral signs of tinnitus. This was a complex, long lasting, and dynamic process. On the day following exposure, all mice demonstrated signs of acute tinnitus over the entire range of sound frequencies used for testing (10-31 kHz). However, 2-3 months later, a behavioral evidence of tinnitus was evident only at a narrow frequency range (20-31 kHz) representing a presumed chronic condition. Extracellular recordings confirmed a significantly higher rate of spontaneous activity in inferior colliculus neurons in sound-exposed compared to control mice. Surprisingly, unilateral sound exposure suppresses startle responses in mice and they remained suppressed even 3 months post-exposure, whereas auditory brainstem response thresholds were completely recovered during 2 months following exposure. In summary, behavioral evidence of tinnitus can be reliably developed in mice by sound exposure, and tinnitus induction can be assessed by quantifying prepulse inhibition of the acoustic startle reflex.