Systems and methods for detecting the presence of electrically evoked compound action potentials (eCAP), estimating survival of auditory nerve fibers, and determining effects of advanced age on the electrode-neuron interface in cochlear implant users

Abstract

Disclosed herein are of systems, methods, and computer-program products for determining if a response is an electrically evoked compound action potential (eCAP), refining raw data of an eCAP amplitude growth function (AGF) and utilizing maximum (i.e., steepest) slope from moving linear regression to effectively estimate cochlear nerve function, and determining quality of an electrode-neuron interface (ENI) using a model developed from eCAP attributes.

Description

The eCAP RRF was obtained with two biphasic, charge balanced, electrical pulses using a modified template subtraction method (Miller et al., 2000). The masker pulse was presented at C level and the probe pulse was presented at 10 CLs below C level. eCAPs were recorded as the MPI was systematically increased from 100 ps to 10 ms. The maximum masker stimulation level for the eCAP RRF was the same as the masker stimulation level for the eCAP I / O function for all except three electrodes (A2, electrode 18; A8, electrodes 12 and 21). For these three electrodes, there was a 1 CL difference between the maximum masker stimulation levels used to measure these two functions. Speech Perception Scores Participants’ speech perception capabilities were evaluated using ConsonantNucleus-Consonant (CNC) word lists (Peterson & Lehiste, 1962) and AzBio sentences (Spahr et al., 2015) in quiet. Results of several studies have shown that cognitive function plays an important role for speech perception in noise (e.g., Dryden et al., 2017; Nuesse et al., 2018). Therefore, speech perception scores were only measured in quiet in this study in order to minimize the effects of cognitive function on study results. All speech perception testing took place in sound-proof booths, using the procedure described in the new Minimum Speech Test Battery (MSTB, 2011). The auditory stimuli were presented in the sound booth via a speaker placed one meter in front of the participant at zero degrees azimuth, calibrated to 60 dB(A) sound pressure level using a sound level meter. For the four participants who are bilateral Cl users (A4, A5, A7 and A13), speech perception scores were measured for each test ear separately. Data Analysis eCAP Refractory Recovery Function The top panels of Figure 1 show eCAPs recorded at different MPIs for electrical stimulations at electrode 3 in participants CND22, NSCN9, and A12, respectively. The eCAP amplitudes measured at different MPIs were normalized to the eCAP amplitude measured at 10 ms and plotted as a function of MPI to generate the eCAP RRF. Two children with CND (CND5 and CND16) did not have a recorded eCAP with an MPI of 10 ms at the apical electrode, and so the eCAP amplitude recorded at the next longest MPI (i.e., 8.1 ms) was used for normalization. The bottom panels of Figure 1 show the normalized eCAP amplitudes along with the fitted exponential decay functions used to estimate ARP and RRP, at electrode 3 in participants CND22, NSCN9, and A12, respectively. For all participants, estimates of the ARP (i.e., to) and RRP (i.e., t) were found using statistical modeling with the exponential decay function MPI-t0 eCAPN = A - Ae ~        (1) where eCAPN is the normalized eCAP amplitude, A represents the maximum normalized eCAP amplitude, and MPI is the masker probe interval in ms. This exponential decay function has been used to create the eCAP RRF and estimate the ARP and RRP in previously published studies (e.g., Morsnowski et al., 2006; Botros & Psarros, 2010; Wiemes et al., 2016; He et al., 2018). When the ARP estimate was unreasonable (i.e., t0 < 0) due to data recordings that were poorly represented by the exponential decay function, the ARP was estimated as the shortest MPI that was longer than 350 ps at which an eCAP was recorded. The shortest MPI was used instead of t0 in 10% of the electrodes tested. Poor fitting of the exponential function occurred most frequently in children with CND. Substitution of shortest MPI for t0 was only performed on four electrodes for participants who did not have CND (A13L, electrode 3; A14, electrode 3; NSCN27, electrodes 3 and 12). Insert Figure 1 about here eCAP I / O function Stimulation levels were converted to units of electrical charge in nanocoulombs (nC) per phase due to variations in PPD among participants. The top panels of Figure 2 show eCAPs recorded at different stimulation levels at electrode 21 in participants CND22, NSCN2, and A11, respectively. The bottom panels of Figure 2 show eCAP amplitudes as a function of stimulation level. The slope of the eCAP I / O function was estimated using linear regression with the linear function eCAP = a * SL + b              (2) where eCAP is the eCAP amplitude in pV, a represents the slope of the eCAP I / O function, SL is the stimulation level in nC, and b represents the intercept of the function with the vertical axis. Linear regression is the most commonly used function to estimate the slope of the eCAP I / O function (e.g., Brown et al., 1990; Kim et al., 2010; Schvartz-Leyzak and Pfingst, 2016). Insert Figure 2 about here Predictive models Model variables eCAP parameters used as input variables in the predictive models were derived from the eCAP RRF and eCAP I / O function. These parameters include t0, the eCAP threshold, slope of the eCAP I / O function, and N1 latency of the eCAP with the maximum amplitude. The output variable is a number which represents the functional status of the CN. For model training, the value of this variable was 0 for children with CND and 1 for children with NSCNs. For model prediction, the output is a value between 0 and 100, where 0 represents the poorest neural function among study participants and 100 represents the best neural function among participants included in the study. This number between 0 and 100 is defined as the CN index. While the relationship between CN neural function and N1 latency has not been well studied in the literature, we have observed that eCAPs recorded in children with CND have prolonged N1 latencies compared to those recorded in children with NSCNs (Xu et al., 2019). Additionally, independent two-sample t-tests comparing the CND and NSCN study groups revealed significant differences for all of the included eCAP parameters: t0 (t<i66) = 6.15, p<0.001), the eCAP threshold (t(iee) = 15.89, p<0.001), slope of the I / O function (t(iee) = 9.06, p<0.001), and N1 latency (t(iee) = 12.11, p<0.001). Therefore, all of these eCAP parameters were included in the predictive models. The eCAP amplitude measured at C level was not included in the models because it is strongly correlated with the slope of the eCAP I / O function (r=0.86, p<0.001) and provides redundant information. t and P2 latency were not included in the models because other studies suggest that t is not related to CN function (Fulmer et al., 2010; Lee et al., 2012; Wiemes et al., 2016; He et al., 2018), and P2 latency is statistically dependent on N1 latency. The eCAP parameters recorded at each electrode site were included together in one combined vector because the aim of this study was to create an objective model that predicts overall CN function for individual patients. The eCAP parameters were concatenated based on known patterns of neural function (i.e., “low”, “medium”, “high”) to provide a consistent comparison across patient populations for estimating overall CN function. Specifically, the eCAP parameters were concatenated from basal to apical electrode site (i.e., “basal”, “middle”, “apical”) for adult and NSCN groups. The eCAP parameters for the CND group were arranged in reverse order (i.e., “apical”, “middle”, “basal”) because children with CND have better neural function in the basal region compared to the apical region (He et al., 2018), which is opposite to the neural-functional pattern of typical Cl users (Propst et al., 2006; Gordon et al., 2007; Brill et al., 2009; Hughes et al., 2009). Model structure eCAP parameters in children with CND and children with NSCNs were used as the training dataset for regression models that separate CN function between these two patient populations. Specifically, each eCAP parameter was standardized across all pediatric participants to eliminate any bias due to differences in scale between the eCAP parameters. Each eCAP parameter was standardized according to x =                (3) where x is a vector containing the normalized value for the eCAP parameter for each pediatric participant, x' is a vector containing the non-normalized value for the eCAP parameter for each pediatric participant, and / t and a are the is the mean and standard deviation of x', respectively. The recorded eCAP parameters for each adult participant were also standardized using the means and standard deviations from the pediatric (i.e., training) data according to Equation 3 for model prediction. The twelve standardized eCAP parameters (4 eCAP parameters x 3 electrode locations) from the pediatric participants were used as the input variables (xr,x2, ...,x12), to train the predictive models. The output variable (y) used for model training was determined by study group, where y = 0 for children with CND and y = 1 for children with NSCNs. The model parameters ..., / ?i2) were found by performing regression analyses with three supervised machine learning algorithms: linear regression with elastic net regularization, SVM regression with a linear kernel, and logistic regression with elastic net regularization. The mathematical formulation for each regression model is presented in Table 2. Insert Table 2 about here Elastic net regularization was used for linear and logistic regression because it improves the accuracy of model predictions by preventing overfitting of the model to the training data. Moreover, elastic net regularization performs variable selection which produces a sparse model for improved interpretability of the model structure. SVM regression has L2 norm regularization built into its default algorithm. The hyperparameters used to find the model parameters were selected by minimizing the mean square prediction error estimated through five-fold cross validation. Once the model parameters were found, the standardized eCAP data from each participant were mapped through the model function to obtain a predicted output variable (yp) for each participant. For linear and SVM regression, yp = 0Tx + 0O, where = [^1 ^2 ■" ^12]> X = [%1,x2, ...,x12]T, and T represents the vector transpose. For logistic regression, yp = (1 + e pTx . Finally, the output variable was scaled into the interval [0, 100] to create the CN index. The CN index was calculated for each participant from the predicted output variable from linear and SVM regression as CN index = 100 * (yp - ymin) / (ymax - ymin)> where ymax and ymin are the maximum and minimum predicted output variables across all participants, respectively. For logistic regression, CN index = 100 * yp. Statistical Analysis Statistical modeling and analysis for this study was performed using MATLAB (Mathworks Inc., version 2019b) software. The trust-region-reflective algorithm was used to estimate parameters of the mathematical functions used in statistical modeling. The one-way analysis of variance (ANOVA) with the Tukey’s honest significant difference (HSD) post-hoc test was used to compare each eCAP parameter among study groups and across electrode locations. ANOVA and Tukey’s HSD criterion were also used to compare CN indices across study groups. One-tailed Pearson correlation analysis was used to evaluate the association between CN indices and CNC word scores measured in adult participants. The one-tailed Spearman rank correlation test was used to evaluate the association between CN indices and AzBio sentence scores measured in adult participants because the AzBio sentence scores were not normally distributed. All statistical analyses were performed at the 95% confidence level. RESULTS eCAP Parameters The mean and standard deviations of eCAP parameters used in the models for all three study groups recorded at three electrode locations are shown in Figure 3. As observed in the figure, the CND group has higher t0 and eCAP threshold values, smaller slopes of the I / O function, and longer N1 latencies at all electrode locations when compared to the other two study groups. There was a significant difference in these eCAP parameters among study groups at all electrode locations (F(2,77)£ 5.62, p<0.005). Statistically significant post-hoc comparisons between study groups are indicated with asterisks in Figure 3. Insert Figure 3 about here Another observed trend for the CND group is that t0 and threshold values tend to increase as the electrode location moves from the basal to more apical regions of the cochlea. Statistical analyses confirmed significantly larger t0 (p=0.002) and higher threshold (p=0.002) values recorded at the apical electrode site than at the basal electrode site for children with CND. Trends that existed for the adult group included decreasing sample means of t0 and N1 latency, as the electrode site moved from the base to the apex. In agreement with these observed trends, t0 values were significantly smaller and N1 latencies were significantly shorter at the apical electrode than at the basal electrode (p=0.002 and p=0.007, respectively) for the adult Cl users. No trend in the data was readily observed for the NSCN group. Details of statistical findings of each study group when comparing the eCAP parameters across electrode locations are listed in Table 3. Insert Table 3 about here Model Structure The model parameters for each of the three predictive models are provided in Table 4. Each regression algorithm found model parameters which were substantially different from one another. However, 02 was the largest in magnitude for all three models, when excluding the model intercept term ( / ?0). Insert Table 4 about here CN Index The CN index calculated using each predictive model is shown for each participant in Figure 4. A line is drawn to indicate results of each participant across models. First of all, it is clearly seen that CN indices for all children with CND, with one notable exception (CND12), is smaller than CN indices for children with NSCNs for each model. It is also apparent that the adult participants, as a group, had CN indices that are comparable to the children with NSCNs and greater than children with CND. There was a significant difference in CN index between study groups for results of all models (Linear: F<2,77) = 136.11, p<0.001; SVM: F(2,77) = 136.12, p<0.001; Logistic: F(2,77) = 534.23, p<0.001). Results from multiple comparisons using Tukey’s HSD criterion showed that CN indices for children with CND were significantly smaller than CN indices for children with NSCNs and adult participants for results of all three models (p<0.001 for all comparisons). There was not a significant difference between the adult group and children with NSCNs for CN indices calculated using any of the models (Linear: p=0.91, SVM: p=0.99, Logistic: p=0.91). Insert Figure 4 about here We also observe that the relative order of CN index among individual participants within each study group (i.e., rank) is generally consistent across models. While some lines cross each other, CN indices calculated using different models are generally in the same region. The change in individual rank between models, averaged across each group of study participants, was 2.25 (SD: 1.70) for children with CND, 5.19 (SD: 3.48) for children with NSCNs, and 2.75 (SD: 1.96) for adults. Speech Perception Scores Figure 5 shows the results of speech perception tests as a function of CN index calculated using each of the three models for all adults who participated in the study. Results of correlation analyses are included in the lower right-hand corner of each panel. Overall, correlation coefficients ranged from 0.49 to 0.73, showing that higher CN indices were associated with better performance on the speech perception tests. Results of each of the correlation tests were statistically significant and were similar for CN indices calculated using different models. Insert Figure 5 about here DISCUSSION eCAP Parameters Results of this study showed that children with CND had significantly longer absolute refractory periods, higher eCAP thresholds, flatter slopes of I / O functions, and longer N1 latencies than children with normal-size CNs. These results are consistent with those reported in He et al. (2018, 2019a) and Xu et al. (2019). Model Structure Weighting Coefficients The relative magnitude of standardized regression coefficients can be used as a measure of the importance of each input variable in predicting the output variable (Mehmood et al., 2010). Therefore, the magnitudes of the model parameters that scale the eCAP parameters (i.e., & - / ?12) represent the relative importance of each eCAP parameter in creating the CN index. As seen in Table 4, the regression coefficient for the eCAP threshold at the electrode location with the lowest level of expected neural function (i.e., 02) had the highest magnitude among all other regression coefficients (excluding the offset term / ?0) in all three models. This suggests that the eCAP threshold is an important indicator for CN function, especially in regions of poorer neural function. This expectation is supported by an animal study which showed that rats with smaller densities of SGNs had higher response thresholds than normal control rats (Shepherd et. al, 2004). However, Pfingst et al. (2015b) presented results that eCAP thresholds did not predict neural survival in five guinea pigs with various degrees of SGN densities. Factors accounting for the discrepancy in these study results are unclear but may include small sample sizes tested in Pfingst et al. (2015b) and differences in species tested among studies. Machine Learning Algorithms This study utilized the supervised machine learning algorithms of linear regression, logistic regression, and SVM regression. These techniques have been found to be useful in predicting Cl outcomes (Tan et al., 2015; Ramos-Miguel et al., 2015; Guerra-Jimenez et al., 2016; Feng et al., 2018). In this study, machine learning techniques were used to create predictive models for estimating the functional status of the CN. Each algorithm produced CN indices for the adult study group that were significantly correlated with speech perception scores (Figure 5). Additionally, the relative ranking of CN index between participants of all study groups was consistent across models (Figure 4). This consistency provides strong support for the validity of the overall concept and the robustness of the approach used in this study. While not the primary focus of this study, the machine learning algorithms utilized in this study can also be used as classification algorithms, with minor modification (see Text, Supplemental Digital Content 1, which describes the methodology). Each algorithm performs very well with classification accuracies of 91 -95% (see Table A1, Supplemental Digital Content 1, which details the performance of each machine learning algorithm in classifying children with CND and children with NSCNs). This accuracy is comparable with the best classification algorithms reported for Cl studies which range from 49-94% (Tan et al., 2015; Ramos-Miguel et al., 2015; Guerra-Jimenez et al., 2016; Feng et al., 2018). CN Index Pediatric Study Groups This study tested the hypothesis that the predictive models created in this study would accurately stratify individual patients based on the functional status of the CN. We expected that distributions of CN index for children with CND and children with NSCNs would be distinct. Our results showed a clear separation in CN index between these two participant groups regardless of the machine learning algorithm used. These results were consistent with the study hypothesis and followed the expected data trend. A much larger range of CN functional statuses were predicted by the linear and SVM regression model for children with CND compared to children with NSCNs, as shown by the CN index values (Figure 4). The predicted status for children with CND ranged from very poor to good functional status. This result agrees with studies that have reported large ranges in Cl outcomes for children with CND (e.g., Young et al., 2012; Vincenti et al., 2014; Birman et al., 2016; Han et al., 2019). Specifically, some children with CND had no awareness of environmental sounds with Cis, while a few patients could understand speech without visual cues (Young et al., 2012; Vincenti et al., 2014; Birman et al., 2016; Han et al., 2019). Considering that CND is likely caused by arrested inner ear development during embryogenesis (Jackler et al. 1987), varying degrees of development of the CN would be expected depending on when the inner ear stops developing. This is supported by a recent study which reported large differences in the number of electrodes with recordable eCAPs in children with CND (He et al., 2018). Results of that study showed that some children with CND had recordable eCAPs at all electrode locations while eCAPs could not be measured at any electrode location in several patients tested in that same study. In contrast, eCAPs were recorded at all electrode locations for all children with NSCNs. Therefore, a wider range of CN functional status for children with CND compared to children with NSCNs, as estimated by the CN index in this study, would be expected. One patient (CND12) had a predicted functional status within the range of children with NSCNs. Therefore, this particular patient would be expected to have outcomes similar to typical Cl users. Supporting this expectation, this particular patient has developed open-set speech skills. We are currently following up with other pediatric participants tested in this study for their speech and language skill development. Adult Study Group For adult study participants, we expected that CN indices would be generally smaller than those for children with NSCNs but greater than those for children with CND. As we expected, the results showed that the adults had significantly greater CN index values than children with CND. However, the average CN index for the adult study group was not significantly different from the average CN index for children with NSCNs. The lack of statistical significance may be due to high CN indices measured in young adult Cl users. As a preliminary investigation into the relationship between CN function and age, we compared CN indices between the youngest and oldest adults. Specifically, CN indices were compared for the five youngest adults (Participants: A3, A7, A9, A10, and A15; Mean age attesting: 44.16, SD 11.48 years), the five oldest adults (Participants: A5, A14, A16, A17, and A19; Mean age attesting: 78.52, SD 2.61 years) and all of the children with NSCNs. Participants A5 and A7 are bilateral Cl users, so a total of six ears were included in both the youngest and oldest adult groups for these comparisons. The means and standard deviations of CN indices calculated from each of the predictive models are shown in Figure 6 for children with NSCNs, the youngest adult participants, and the oldest adult participants. Insert Figure 6 about here As observed in the figure, the mean CN index value was similar for the youngest adults and the children with NSCNs, both of which were larger than that of the oldest adults for results of all three models. There was a significant difference in CN index among these groups for results of all three models (Linear: F(2,4i) = 4.44, p=0.018, SVM: F(2,41) = 3.61, p=0.036, logistic F(2,41)= 4.93, p=0.012). Multiple comparisons with Tukey’s HSD criterion indicated that the oldest adult group had statistically smaller CN indices than children with NSCNsforthe linear (p=0.017) and logistic (p=0.001) models, but not for the SVM model (p=0.053). The youngest adults had statistically greater CN indices than the oldest adults for the SVM model (p=0.046), but not for the linear or logistic model (p=0.050 and p=0.055, respectively). Finally, there was no statistical difference in CN index between the youngest adult group and children with NSCNs for results of any of the predictive models (Linear: p=0.962, SVM: p=0.702, Logistic: p=1.000). The results of this exploratory investigation supports the idea that the wide range of age at testing (28.73 - 88.80 years) of adult participants may partially account for the non-significant difference in CN indices between the NSCN and the adult study group. Moreover, these data suggest that older patients have worse CN function (as indicated by smaller CN indices), which agrees with other studies showing deteriorating CN function with advanced age (e.g., McFadden et al., 1997; Makary et al., 2011; Viana et al., 2015; Wu et al., 2019). A comprehensive analysis of the effect of aging on CN function is currently under investigation as a separate study. Nevertheless, the present investigation provides additional support for the idea that the CN index developed in this study reflects CN functional status. We also expected that CN index would be positively correlated with speech outcome measures. Confirming this expectation, CN index was positively and significantly correlated with speech perception of CNC words and AzBio sentences in quiet (Figure 5). Significant correlations were observed for CN indices calculated using each predictive model. This result supports the idea that the CN index represents overall CN function because Cl outcomes are related to CN function (Kim et al., 2010; Zhou & Pfingst, 2014; Schvartz-Leyzac & Pfingst, 2018; He et al., 2018). Furthermore, this result shows the utility and benefit of employing machine learning approaches to predict Cl outcomes, which may have implications for patient treatment and counseling. Study Limitations One potential study limitation is the assumption that patterns of neural function for an individual patient follow the trend of its patient population. Specifically, the CN index is derived with the assumption that children with CND have better CN function in basal regions compared to apical regions, and vice versa for children with NSCNs and adult patients. While the literature confirms these overall trends (e.g., Propst et al., 2006; Gordon et al., 2007; Brill et al., 2009; Hughes et al., 2009; He et al., 2018), individual variations exist. Any deviations from the expected pattern will affect the CN index calculation. This limitation could be eliminated with a future optimized model that predicts CN function at individual electrode sites. Another potential limitation is that the eCAP is a neurophysiological response that depends on a sufficient number of CN fibers responding synchronously to electrical stimulations. Therefore, it is a composite measure of overall CN neural survival and the integrity of individual CN fibers. As such, it is not known if poorer eCAP responses are due to a decreased number of CN fibers or due to degeneration of existing CN fibers. Moreover, this may be different across patient populations. For example, eCAP responses might primarily be affected by the few number of intact CN fibers for children with CND, whereas, children with NSCNs and adult patients may have a large number of degenerated CN fibers. Currently, the effect of various pathologies on eCAP responses is not well understood. Nevertheless, the eCAP is a useful response for characterizing the functional status of the CN. Finally, the models created in this study are not able to predict CN function at individual electrode sites. Rather, the models presented in this study predict overall CN function across multiple electrodes. We are currently investigating methods for predicting neural function at individual electrode sites, which may be useful in programming individual patient Cl settings. CONCLUSIONS This study presented models created using three supervised machine learning techniques that generate an index for the functional status of the CN based on eCAP recordings for individual patients. All three models successfully stratified Cl patients based on their CN functional statuses. Specifically, children with NSCNs had significantly better predicted CN functions than children with CND. Adult Cl users had a range of predicted CN functions that were positively and significantly correlated with scores on speech perception tests. Results of this study suggested that these models may be useful for developing objective clinical tools for optimizing Cl programming settings and predicting Cl outcomes for individual Cl patients. ACKNOWLEDGMENTS We gratefully thank all participants and their families for engaging in this study. We also thank the three anonymous reviewers for their insightful comments and suggestions. REFERENCES Abbas, P. J., Brown, C. J., Shallop, J. K., et al. (1999). Summary of results using the nucleus CI24M implant to record the electrically evoked compound action potential. Ear Hear, 20, 45-59. Abbas, P. J., Hughes, M., Brown, C. J., et al. (2004). Channel Interaction in Cochlear Implant Users Evaluated Using the Electrically Evoked Compound Action Potential. Audiol Neurootol. 9. 203-13. Adunka, O. F., Roush, P. A., Teagle, H. F., et al. (2006). Internal auditory canal morphology in children with cochlear nerve deficiency. Otol Neurotol, 27, 793-801. Beyea, J. A., Mcmullen, K. P., Harris, M. S., et al. (2016). Cochlear Implants in Adults. Otol Neurotol, 37, 1238-1245. Birman, C. S., Powell, H. R., Gibson, W. P., et al. (2016). Cochlear implant outcomes in cochlea nerve aplasia and hypoplasia. Otol Neurotol, 37, 438-445. Blarney, P., Artieres, F., Ba§kent, D., etal. (2013). Factors affecting auditory performance of postlinguistically deaf adults using cochlear implants: an update with 2251 patients. Audiol Neurootol, 18(1), 36-47. Botros, A., Psarros, C. (2010). Neural Response Telemetry Reconsidered: II. The Influence of Neural Population on the ECAP Recovery Function and Refractoriness. Ear Hear, 31, 380-391. Botros, A., van Dijk., B., Killian M. (2007). AutoNR: An automated system that measures ECAP thresholds with the Nucleus Freedom cochlear implant via machine intelligence. Artif Intell Med. 40:15-28. Brill, S., Muller, J., Hagen, R., et al. (2009). Site of cochlear stimulation and its effect on electrically evoked compound action potentials using the MED-EL standard electrode array. BioMed Eng OnLine 8, 40. Brown, C. J., Abbas, P. J., Gantz, B. (1990). Electrically evoked whole-nerve action potentials: data from human cochlear implant users. J Acoust SocAm, 88, 1385-1391. Brown, C. J., Abbas, P. J., Gantz, B. (1998). Preliminary experience with Neural Response Telemetry in the Nucleus CI24M cochlear implant. Am. J. Otol. 19, 320-327. Buchman, C. A., Roush, P. A., Teagle, H. F., et al. (2006). Auditory neuropathy characteristics in children with cochlear nerve deficiency. Ear Hear, 27, 399-408. Clemmens, C. S., Guidi, J., Caroff, A., et al. (2013). Unilateral cochlear nerve deficiency in children. Otolaryngol Head Neck Surg, 149, 318-325. Crowson, M. G., Lin, V., Chen, J. M., & Chan, T. C. (2020). Machine Learning and Cochlear Implantation-A Structured Review of Opportunities and Challenges. Otol Neurotol. 41(1), e36-e45. Dryden, A., Allen, H.A., Henshaw, H., et al. (2017). The association between cognitive performance and speech-in-noise perception for adult listeners: a systematic literature review and meta-analysis. Trend Hear, 21, 1-21. Feng, G., Ingvalson, E. M., Grieco-Calub, T. M., et al. (2018). Neural preservation underlies speech improvement from auditory deprivation in young cochlear implant recipients. Proc Natl Acad Sci USA. 115:E1022-31. Firszt J. B., Holden L. K., Skinner M. W., et al. (2004). Recognition of speech presented at soft to loud levels by adult cochlear implant recipients of three cochlear implant systems. Ear Hear, 25, 375-87. Fulmer, S. L., Runge, C.L., Jensen, J.W., et al. (2010). Rate of Neural Recovery in Implanted Children with Auditory Neuropathy Spectrum Disorder. JAMA Otolaryngol Head Neck Surg, 144, 274-279. Garadat, S. N., Zwolan, T. A., & Pfingst, B. E. (2012). Across-site patterns of modulation detection: relation to speech recognition. J Acoust SocAm, 131(5), 4030-4041. Gartner, L., Lenarz, T., Joseph, G., Buchner, A. (2010). Clinical use of a system for the automated recording and analysis of electrically evoked compound action potentials (ECAPs) in cochlear implant patients. Acta Otolaryngol. 130:724-32. Glastonbury, C. M., Davidson, H. C., Harnsberger, H. R., Butler, J., Kertesz, T. R. and Shelton, C. (2002). Imaging findings of cochlear nerve deficiency. AJNR Am J Neuroradiol, 23(4), pp.635-643. Gordon, K., Papsin, B., Harrison, R. (2007). Auditory brainstem activity and development evoked by apical versus basal cochlear implant electrode stimulation in children. Clin Neurophysiol, 118, 1671-1684. Guerra-Jimenez, G., Ramos De Miguel, A., Falcon Gonzalez, J. C. et al. (2016). Cochlear implant evaluation: Prognosis estimation by data mining system. J Int Adv Otol. 12:1-7. Han, J. J., Suh, M. W., Park, M. K., et al. (2019). A Predictive Model for Cochlear Implant Outcome in Children with Cochlear Nerve Deficiency. Sci Rep, 9(1), 1154. He, S., Shahsavarani, B. S., McFayden, T. C., et al. (2018). Responsiveness of the electrically stimulated cochlear nerve in children with cochlear nerve deficiency. Ear Hear, 39, 238-250. He, S., Xu, L., Skidmore, J., et al. (2019a). The effect of interphase gap on neural response of the electrically-stimulated cochlear nerve in children with cochlear nerve deficiency and children with normal-sized cochlear nerves. Ear Hear, Oct. 31 (Epub ahead of print). He, S., Chao, X.H., Wang, R.J., et al. (2019b). Recommendations for measuring the electrically evoked compound action potential in children with cochlear nerve deficiency. Ear Hear, Sep. 13 (Epub ahead of print). He, S., Teagle, H.F.B., Buchman, C. (2017). The electrically evoked compound action potential: from laboratory to clinic. Front Neurosci, 11:339. Holden, L.K., Finley, C.C., Firszt, J.B., et al. (2013). Factors affecting open-set word recognition in adults with cochlearimplants. Ear Hear, 34, 342-360. Hughes, M. L., & Stille, L. J. (2009). Psychophysical and physiological measures of electrical-field interaction in cochlearimplants. J. Acoust. Soc. Am., 125(^), 247-260. Jackler, R. K., Luxford, W. M., House, W. F. (1987). Congenital malformations of the inner ear: a classification based on embryogenesis. Laryngoscope, 97(3 Pt 2; Suppl 40), 2-14. Kang, W. S., Lee, J. H., Lee, H. N., & Lee, K. S. (2010). Cochlearimplantations in young children with cochlear nerve deficiency diagnosed by MRI. Otolaryngology—Head and Neck Surgery, 143(^), 101-108. Kim, J.R., Abbas, P.J., Brown, C.J., et al. (2010). The relationship between electrically evoked compound action potential and speech perception: a study in cochlear implant users with short electrode array. Otol Neurotol, 31, 1041-1048. Kirby, A. E., & Middlebrooks, J. C. (2010). Auditory temporal acuity probed with cochlear implant stimulation and cortical recording. J Neurophysiol, 103(^), 531-542. Kirby, A. E., & Middlebrooks, J. C. (2012). Unanesthetized auditory cortex exhibits multiple codes for gaps in cochlear implant pulse trains. J Assoc Res Otolaryngol, 13(^), 67-80. Kutz, J. W. Jr, Lee, K. H., Isaacson, B., et al. (2011). Cochlear implantation in children with cochlear nerve absence or deficiency. Otol Neurotol, 32, 956-961. Lazard, D.S., Vincent, C., Venail, F., et al. (2012). Pre-, per- and postoperative factors affecting performance of postlinguistically deaf adults using cochlear implants: a new conceptual model overtime. PLoS One, 7, e48739. Lee, E.R., Friedland, D.R., Runge, C.L. (2012). Recovery From Forward Masking in Elderly Cochlear Implant Users. Otol Neurotol, 33, 355-363. Long, C. J., Holden, T. A., McClelland, G. H., et al. (2014). Examining the electro-neural interface of cochlear implant users using psychophysics, CT scans, and speech understanding. J Assoc Res Otolaryngol, 15(2), 293-304. Luo, J.F., Xu, L., Chao, X.H., et al. (2019). The effects of GJB2 and SLC26A4 gene mutations on neural response of the electrically-stimulated auditory nerve in children. Ear Hear, (in press). Makary, C. A., Shin, J., Kujawa, S. G., et al. (2011) Age-Related Primary Cochlear Neuronal Degeneration in Human Temporal Bones. JARO 12, 711-717. McFadden, S .L., Campo, P., Quaranta, N., etal. (1997). Age-related decline of auditory function in the chinchilla (Chinchilla laniger). Hear. Res., 111, 114-126. Mehmood, T., Liland, K. H., Snipen, L., et al. (2012). A review of variable selection methods in Partial Least Squares Regression. Chemometrics and Intelligent Laboratory Systems, 118, 62-69. Miller, C. A., Abbas, P. J., Brown, C. J. (2000). An Improved Method of Reducing Stimulus Artifact in the Electrically Evoked Whole-Nerve Potential. Ear Hear, 21,280-290. Moberly, A.C., Bates, C., Harris, M.S., et al. (2016). The Enigma of Poor Performance by Adults with Cochlear Implants. Otol Neurotol, 37, 1522-1528. Morsnowski, A., Charasse, B., Collet, L., et al. (2006). Measuring the Refractoriness of the Electrically Stimulated Auditory Nerve. Audiol Neurootol, 11, 389-402. MSTB (2011). Minimum speech test battery for adult cochlear implant users. http: / / auditorypotential.com / MSTBfiles / MSTBManual2011-06-20.pdf. Accessed Nov 12, 2019. Nuesse, T., Steenken, R., Neher, T., et al. (2018). Exploring the link between cognitive abilities and speech recognition in the elderly under different listening conditions. Front Psychol, 9, 678. Peterson, G. E., & Lehiste, I. (1962). Revised CNC lists for auditory tests. J Speech Hear Disord., 27(1), 62-70. Pfingst, B.E., Colesa, D.J., Watts, M.M., et al. (2017). Neurotrophin gene therapy in deafened ears with cochlear implants: long-term effects on nerve survival and functional measures. J Assoc Res Otolaryngol, 18, 731-750. Pfingst, B.E., Hughes, A.P., Colesa, D.J., et al. (2015a). Insertion trauma and recovery of function after cochlear implantation: evidence from objective functional measures. Hear Res, 330, 98-105. Pfingst, B. E., Zhou, N., Colesa, D. J., et al. (2015b). Importance of cochlear health for implant function. Hear Res, 322, 77-88. Prado-Guitierrez, P., Fewster, L.M., Heasman, J.M., et al. (2006). Effect of interphase gap and pulse duration on electrically evoked potentials is correlated with auditory nerve survival. Hear Res, 215, 47-55. Propst, E.J., Papsin, B.C., Stockley, T.L., et al. (2006). Auditory responses in cochlear implant users with and without GJB2 deafness. Laryngoscope, 116, 317-327. Ramekers, D., Versnel, H., Strahl, S.B., et al. (2014). Auditory-nerve response to varied inter-phase gap and phase duration of the electric pulse stimulus as predicators for neuronal degeneration. J Assoc Res Otolaryngol, 15, 187-202. Ramos-Miguel, A., Perez-Zaballos, T., Perez, D., et al. (2015). Use of data mining to predict significant factors and benefits of bilateral cochlear implantation. Eur Arch Otorhinolaryngol. 272:3157-62. Schvartz-Leyzac, K. C., Pfingst, B. E. (2018). Assessing the relationship between the electrically evoked compound action potential and speech recognition abilities in bilateral cochlear implant recipients. Ear Hear, 39, 344-358. Shepherd, R. K., Roberts, L. A., Paolini, A. G. (2004). Long-term sensorineural hearing loss induces functional changes in the rat auditory nerve. EurJ Neurosci, 20, 3131-3140. Spahr, A. J., Dorman, M. F., Litvak, L. M., et al. (2012). Development and validation of the AzBio sentence lists. Ear Hear, 33(1), 112. Tan, L., Holland, S. K., Deshpande, A. K., etal. (2015). A semi-supervised Support Vector Machine model for predicting the language outcomes following cochlear implantation based on preimplant brain fMRI imaging. Brain Behav. 5:1-25. Teagle, H. F., Roush, P. A., Woodard, J. S., et al. (2010). Cochlear implantation in children with auditory neuropathy spectrum disorder. Ear Hear, 31, 325-335. Van de Heyning, P., Arauz, S. L., Atlas, M., et al. (2016). Electrically evoked compound action potentials are different depending on the site of cochlear stimulation. Cochlear Implants I nt, 17, 251-262. van Dijk, B., Botros, A. M., Battmer R. D., et al. (2007). Clinical results of AutoNRT™, a completely automatic ECAP recording system for cochlear implants. Ear Hear. 28:55870. Viana, L. M., O'Malley, J. T., Burgess, B. J., et al. (2015). Cochlear neuropathy in human presbycusis: Confocal analysis of hidden hearing loss in post-mortem tissue. Hear Res, 327, 78-88. Vincenti, V., Ormitti, F., Ventura, E., et al. (2014). Cochlear implantation in children with cochlear nerve deficiency. Int J Pediatr Otorhinolaryngol, 78, 912-917. Wei, X., Li, Y., Chen, B., et al. (2017). Predicting auditory outcomes from radiological imaging in cochlear implant patients with cochlear nerve deficiency. Otol Neurotol, 38(5), 685-693. Wemes, G. R. M., Hamerschmidt, R., Moreira, A. T. R., et al. (2016). Auditory Nerve Recovery Function in Cochlear Implant Surgery with Local Anesthesia and Sedation versus General Anesthesia. Audiol Neurootol, 21, 150-157. Wu, P. Z., Liberman, L. D., Bennett, K., et al. (2019). Primary Neural Degeneration in the Human Cochlea: Evidence for Hidden Hearing Loss in the Aging Ear. Neuroscience, 407, 8-20. Xu, L., Skidmore, J., Luo, J., etal. (2019). The effect of pulse polarity on neural response of the electrically-stimulated cochlear nerve in children with cochlear nerve deficiency and children with normal-sized cochlear nerves. Ear Hear. (Accepted) Young, N. M., Kim, F. M., Ryan, M. E., et al. (2012). Pediatric cochlear implantation of children with eighth nerve deficiency. Int J Pediatr Otorhinolaryngol, 76, 1442-1448. Zekveld, A., Kramer, S.E., & Festen, J. (2011). Cognitive load during speech perception in noise: the influence of age, hearing loss, and cognition on pupil response. Ear Hear, 32, 498-510. Zhou, N., Pfingst, B.E. (2014). Relationship between multipulse integration and speech recognition with cochlear implants. J Acoust See Am, 136, 1257-1268. FIGURES Figure 1. 1170 1450 1797 2227 2759 3420 4239 5253 6510 8069 10000 400  800 1200 1680 2080 CND22, e3 Time (ps) 1450 1797 2227 5253 10000 2759 3420 4239 6510 8069 614 761 944 1170 NSCN9, e3 400 60S 1200 1608 2080 Time (ps) 761 944 1170 1450 1797 2227 2759 3420 4239 5253 6510 8069 50yV 10000 A12, e3 488  380 1300 1600 2000 Time (ps) Upper panels: electrically-evoked compound action potential (eCAP) waveforms measured at different masker-probe intervals (MPIs) for stimulating electrode 3 in one child with cochlear nerve deficiency (CND; CND22), one child with normalsized cochlear nerve (NSCN; NSCN9), and one adult (A12). eCAPs are arranged based on MPI duration, with responses evoked by short MPIs displayed at the top. Each waveform is labeled with the corresponding MPI duration in p.s. Lower panels: refractory recovery functions (round symbols) obtained from the waveforms in the upper panels. The fitted exponential decay function for each refractory recovery function (black line) is also provided. The participant and electrode number are included in the lower right corner of each panel. The stimulations were performed at the maximum comfortable level for each participant and electrode. Figure 2. 22,18 CND22,e21 29.08 29.61 25.63 26,09 26,57 27.05 27.55 28.05 28.56 22.58 22.99 23.41 23.84 24.27 24.72 25.17 430 BOO 1200 1800 2000 Tim® (ys) 403 890 1290 1800 2000 Time (ps) □OyV 400 800 1209 7.19 7.32 7.45 7.59 7.73 7.87 8.01 8,16 8,31 8,46 8.61 9.43 10.32 10.89 11.09 11.29 11.50 11.71 11.92 12.14 12.36 12,58 12.81 13.05 All, ©21 ----r" 1690 2300 Time (ye) 25      30      35                5     19     15    20                5     19     15    29 Stimulation level (nC) Stimulation level (nC) Stimulation level (nC) Upper panels: electrically-evoked compound action potential (eCAP) waveforms measured at different stimulation intensities for electrode 21 in one child with cochlear nerve deficiency (CND; CND22), one child with normal-sized cochlear nerve (NSCN; NSCN2), and one adult (A11). eCAPs are arranged based on stimulation level, with responses evoked by the smallest stimulation level (i.e., eCAP threshold) displayed at the top. Each waveform is labeled with the corresponding probe stimulation level in nanocoulombs (nC). The largest stimulation level presented was the maximum comfortable level (i.e., C level). Lower panels: eCAP input / output functions (round symbols) obtained from the waveforms in the upper panels. The participant and electrode number are included in the lower right corner of each panel. Figure 3. Results of eCAP parameters (mean and standard deviation) measured for three study groups. Ordered in rows from top to bottom are the estimated absolute refractory recovery times (i.e., t0), eCAP thresholds, slopes of eCAP I / O functions, and N1 peak latencies. Results measured at the basal, middle and apical electrode location are provided in the left, middle and right columns, respectively. Statistically significant group comparisons are indicated with asterisks. *: p<0.05, **: p<0.01, ***: p<0.001. Figure 4. CND           NSCN           Adult Lin SWI Log Lin SW1 Log tin SW Log Regression model Results of cochlear nerve (CN) indices obtained from linear (Lin) regression, support vector machine (SVM) regression, and logistic (Log) regression for three study groups. Figure 5. O U £ as 90     189 p = 0.73 p < 0.001 CN index p - 0.62 Ap < 0.001 89     90     190 CM index CM index Results of speech perception of Consonant-Nucleus-Consonant (CNC) words (top row) and AzBio sentences (bottom row) as a function of cochlear nerve (CN) index for all adult participants. Results are displayed for CN indices obtained from linear regression, support vector machine (SVM) regression, and logistic regression. Results of Pearson and Spearman correlation tests are also provided in the bottom right corner of all panels. Figure 6. I-----------------1 Support Vector Machine Regression modei Logistic The means and standard deviations of cochlear nerve (CN) indices calculated from three predictive models for children with normal-sized CNs, the youngest adult participants, and the oldest adult participants. Statistically significant differences (p<0.05) are indicated by an asterisk. TABLES TABLE 1. Demographic information of all subjects who participated in this study Subject Number Sex Ear Tested AAI (yrs) AAT (yrs) Electrodes Tested PPD (MS) Internal Device and Electrode Array MRI Result CN Cochlea CND1 F L 1.2 2.5 1, 5, 10 50, 50, 50 24RE (CA) * IP-2 CND2 M R 2.1 4.1 1, 5, 8 50, 50, 50 24RE (CA) * Normal CND3 M R 6.6 8.5 3, 6, 10 50, 50, 50 24RE (CA) # Normal CND4 F L 1.3 3.3 3, 12, 21 50, 50, 50 24RE (CA) * Normal CND5 F L 3.4 5.4 3, 10, 18 50, 50, 50 24RE (CA) * Normal CND6 M L 1.9 2.4 1, 7, 14 50, 50, 50 24RE (CA) * Normal CND7 M L 4.7 10.8 15, 19, 22 37, 37, 37 24RE (ST) Small IP-2 CND8 F L 1.69 3.81 6, 12, 18 75, 75, 88 24RE (CA) Small Normal CND9 F L 2.1 8.2 3, 11, 16 75, 50, 50 24RE (CA) Absent Normal CND10 F L 2.10 5.50 2, 4, 10 75, 100, 100 24RE (CA) # Normal CND11 M L 7.84 10.03 3, 12, 21 75, 75, 75 24RE (CA) Small Normal CND12 M L 2.10 4.50 3, 12, 17 25, 25, 25 24RE (CA) # Normal °® CND13 M L 9.40 9.71 3, 12, 18 50, 50, 50 24RE (CA) Small Normal CND14 F L 5.10 7.20 3, 9, 15 88, 88, 88 CI512 Small Normal CND15 M L 4.10 7.90 1, 6, 9 50, 50, 50 24RE (CA) Absent Normal CND16 M R 3.76 4.09 1, 6, 9 75, 75, 75 24RE (CA) Small Normal CND17 M R 2.50 4.76 6, 15, 21 50, 50, 50 24RE (CA) # Normal CND18 F R 3.96 4.96 3, 12, 21 37, 37, 37 24RE (CA) * Normal CND19 F R 2.93 6.49 1, 5, 8 50, 50, 50 24RE (CA) * Normal CND20 F R 2.10 4.20 1, 4, 7 50, 50, 50 24RE (CA) Small Normal CND21 F L 1.90 2.50 3, 12, 21 50, 50, 50 24RE (CA) # Narrow CND22 M L 2.10 4.70 3, 12, 21 37, 37, 50 24RE (CA) * Normal CND23L F L 10.83 15.03 3, 6, 9 37, 50, 62 24RE (CA) Small IP-2 CND23R F R 3.84 15.13 3, 12, 21 50, 50, 50 24RE (CA) Small IP-2 NSCN1 F R 2.41 3.00 3, 12, 21 25, 25, 25 CI422 NSCN2 M L 3.52 4.37 4, 12, 21 25, 25, 25 24RE (CA) WO 2021 / 237059                                                 PCT / US2021 / 033604 NSCN3 F R 0.94 2.14 3, 12, 21 25, 25, 25 CI422 NSCN4 M R 2.31 6.78 3, 12, 21 25, 25, 25 24RE (CA) NSCN5 M L 1.57 3.17 3, 12, 21 25, 25, 25 24RE (CA) NSCN6 M R 3.60 6.28 3, 12, 21 25, 25, 25 24RE (CA) NSCN7 F R 3.50 6.50 3, 12, 21 25, 25, 25 CI512 NSCN8 M L 8.50 9.35 3, 12, 21 25, 25, 25 24RE (CA) NSCN9 M L 4.30 8.40 3, 12, 21 25, 25, 25 CI512 NSCN10 F R 3.50 6.50 3, 15, 21 25, 25, 25 CI512 NSCN11 F R 3.00 11.50 3, 12, 21 25, 25, 25 24RE (CA) NSCN12 M L 6.52 7.73 3, 12, 21 25, 25, 25 24RE (CA) NSCN13 F R 2.09 4.3 3, 12, 21 25, 25, 25 24RE (CA) NSCN14 F R 1.65 2.49 3, 12, 21 25, 25, 25 24RE (CA) NSCN15 F L 6.32 8.54 3, 12, 19 25, 25, 25 24RE (CA) NSCN16 M R 0.96 2.91 3, 12, 19 25, 25, 25 24RE (CA) NSCN17 M R 3.33 5.6 3, 12, 19 25, 25, 25 24RE (CA) NSCN18 F L 5.71 8.08 3, 12, 21 25, 25, 25 24RE (CA) NSCN19 M L 2.41 3.41 3, 12, 19 25, 25, 25 24RE (CA) NSCN20 M L 1.66 3.95 3, 12, 19 25, 25, 25 24RE (CA) NSCN21 M R 2.69 3.02 3, 12, 19 25, 25, 25 24RE (CA) NSCN22L M L 1.28 2.64 3, 12, 19 25, 25, 25 CI512 NSCN22R M R 1.28 2.64 3, 12, 19 25, 25, 25 24RE (CA) NSCN23 M R 1.80 3.96 3, 12, 19 25, 25, 25 24RE (CA) NSCN24 M R 6.12 9.83 3, 12, 21 25, 25, 25 24RE (CA) NSCN25L M L 0.96 12.94 3, 12, 21 25, 25, 25 24RE (CA) NSCN25R M R 0.96 12.94 4, 12, 21 25, 25, 25 24RE (CA) NSCN26L M L 2.41 3.41 3, 12, 21 25, 25, 25 24RE (CA) NSCN26R M L 1.66 3.95 4, 12, 21 25, 25, 25 24RE (CA) NSCN27 M R 3.17 6.22 4, 12, 21 25, 25, 25 CI512 NSCN28 M L 1.28 2.64 3, 12, 22 25, 25, 25 Cl 532 NSCN29 F L 0.84 11.94 3, 12, 20 25, 25, 25 24RE (CA) A1 M L 58.85 61.79 3, 12, 20 25, 25, 25 CI512 WO 2021 / 237059                                                 PCT / US2021 / 033604 A2 M L 60.67 69.02 3, 12, 18 25, 25, 25 CI512 A3 M R 43.26 52.69 3, 12, 21 25, 25, 25 24RE (CA) A4L F L 56.01 67.52 3, 12, 21 25, 25, 25 24RE (CA) A4R F R 54.42 67.57 3, 12, 21 25, 25, 25 24RE (CA) A5L M L 72.76 80.80 7, 12, 21 25, 25, 25 CI512 A5R M R 77.54 80.70 3, 9, 21 25, 25, 25 24RE (CA) A6 M R 52.53 61.47 3, 12, 21 25, 25, 25 CI512 A7L F L 54.63 55.36 4, 12, 21 25, 25, 25 Cl 532 A7R F R 44.71 54.73 3, 12, 21 25, 25, 25 24RE (CA) A8 M R 60.31 62.75 4, 12, 21 25, 25, 25 Cl 522 A9 M R 25.60 36.79 3, 12, 21 25, 25, 25 24RE (CA) A10 F L 32.97 36.66 3, 12, 21 25, 25, 25 CI512 A11 F R 48.47 59.6 3, 12, 21 25, 25, 25 24RE (CA) A12 F R 64.92 65.57 3, 12, 21 25, 25, 25 Cl 532 A13L M L 70.21 70.43 3, 12, 21 25, 25, 25 Cl 532 A13R M R 68.67 70.43 3, 12, 21 25, 25, 25 Cl 532 o A14 F L 72.46 76.62 3, 12, 21 25, 25, 25 Cl 522 A15 F R 15.23 28.73 3, 12, 21 25, 25, 25 24RE A16 M R 73.00 74.19 3, 9, 12 25, 25, 25 Cl 522 A17 M L 74.03 79.01 3, 12, 21 25, 25, 25 CI422 A18 M R 58.98 59.52 3, 12, 21 25, 25, 25 Cl 532 A19 F R 75.74 79.78 3, 12, 21 25, 25, 25 24RE A20 M L 68.50 70.23 11, 18, 20 25, 25, 25 Cl 532 *Two Small nerves in the auditory canal #Single nerve in the internal auditory canal (i.e. nerve diameter <3 mm). AAI, age at implantation; AAT, age at testing; PPD, pulse phase duration; CN, cochlear nerve; 24RE (CA), Freedom Contour Advance electrode array; 24RE (ST), Freedom Straight electrode array; IP-2, incomplete partition 2. WO 2021 / 237059                                                 PCT / US2021 / 033604 TABLE 2. Mathematical formulation for three regression models. Model Regularization Optimization problem Hyperparameters Linear Elastic net min w                    / p          v (pTXi + p0- yiy +x[\±p] + \Pj\ ill                                              1       M                   / '=1                      V=1           / . 2 = 0.128 SVM L2 norm min     ^p] + C sz] , such that \(PTXi)a + p0 - y£| < e + Si and s^ > 0 C = 0.001 e = 0.005 a = 2.699 min Logistic Elastic net 2 = 0.013 where SVM, support vector machine; p = [Pt p2... ^i2], vector of model parameters; p0, model intercept term; N = 56, WO 2021 / 237059                                                 PCT / US2021 / 033604 number of observations in training data set; x, vector of electrically-evoked compound action potential (eCAP) parameters; y, output vector; A, regularization parameter; p = 12, number of eCAP parameters; s, vector of slack parameters; C, box constraint; e, error margin; a, kernel scaling factor. tsi WO 2021 / 237059                                                 PCT / US2021 / 033604 TABLE 3. Statistical results when comparing eCAP parameters across electrode locations for three study groups. eCAP Variables Statistical Test CND NSCN Adult to ANOVA F(2,69)=6.59, p=0.002 F(2,93)=4.46, p=0.014 F(2,69)=6.47, p=0.003 HSD B<A, p=0.002 M>A, p = 0.012 B>A, p = 0.002 ANOVA F(2,69)=6.51, p=0.003 F(2,93)=12.40, p<0.001 F(2,69)=6.47, p=0.003 eCAP threshold HSD B<A, p = 0.002 B<M, p<0.001 M>A, p<0.001 B>A, p=0.007 M>A, p=0.001 Slope of I / O ANOVA F(2,69)=0.11, p=0.893 F(2,93)=1.15, p=0.322 F(2,69)=2.20, p=0.118 function HSD NS NS NS N1 latency ANOVA F(2,69)=0.22, p=0.807 F(2,93)=2.91, p=0.060 F(2,69)=5.00, p=0.009 HSD NS NS B>A, p=0.007 ANOVA: analysis of variance; HSD: Tukey’s honest significant difference post-hoc test; CND: cochlear nerve deficiency study group; NSCN: normal-sized cochlear nerve study group; Adult, adult study group; B: basal WO 2021 / 237059                                                 PCT / US2021 / 033604 electrode; M: middle electrode; A: apical electrode; NS: not significant; to: estimate of absolute refractory period derived from the refractory recovery function; eCAP: electrically-evoked Compound Action Potential; I / O: input / output WO 2021 / 237059                                                 PCT / US2021 / 033604 TABLE 4. Model parameters for three predictive models listed by eCAP parameter and the expected neural function at different electrode locations based on literature. Model parameter eCAP parameter Expected _ neural function Regression model Linear SVM Logistic Po 0.571 0.633 -0.584 Pi Low -0.063 -0.038 -0.553 Threshold Low -0.170 -0.040 -2.438 Slope Low 0 -0.001 -0.037 Pl N1 latency Low -0.008 -0.013 0 Po to Medium 0 -0.009 0 Pg Threshold Medium -0.023 -0.019 -1.918 Pi Slope Medium 0 0.008 0 Po N1 latency Medium -0.053 -0.017 -0.055 Po to High 0 -0.008 0 Pio Threshold High 0 -0.014 0 Pll Slope High 0.042 0.011 0 P12 N1 latency High -0.058 -0.024 -0.529 SVM: support vector machine SUPPLEMENTAL DIGITAL CONTENT CLASSIFICATION METHODS We compared three different algorithms for classifying the presence of cochlear nerve deficiency (CND) for pediatric cochlear implant (Cl) users from recordings of electrically-evoked compound action potentials (eCAPs). Each participant was classified as having CND and given a label of 0 or as having a normal-sized cochlear nerve (NSCN) and given a label of 1. The three classifying algorithms that were compared were linear regression with elastic net regularization, support vector machine regression with a linear kernel, and logistic regression with elastic net regularization. The classification boundary was chosen to be 0.5 for all three models. In other words, if the predicted score for an individual patient was greater than 0.5, the individual would be classified as having a NSCN (i.e., label = 1). Otherwise, the individual was classified as having CND (i.e., label = 0). Five-fold cross validation The data from all participants were first randomly split into 5 folds of equal size, each fold having 1 / 5 of the children with CND and 1 / 5 of the children with a NSCN. In each run, 4 folds were used as the training data set to find the model parameters, and the remaining fold was used to test the model. A total of 5 training and testing runs were completed in which each run used a different fold as the testing fold. The model performance was reported as the average over the 5 runs. Evaluation Metrics Accuracy, precision, recall and F-measure score were used to evaluate the performance of the different algorithms and are defined below. For all of these metrics, the possible values are between 0 and 1, where larger values indicate better classification performance. Accuracy: the fraction of all correctly classified participants over all participants. Precision: the fraction of correctly classified children with CND over all the participants that were classified (either correctly or incorrectly) as having CND. Recall: the fraction of correctly classified children with CND over the total number of children with CND. F-measure score: The harmonic mean of precision and recall. Mathematically, this is written as Fl = 2 x-------------). recall+precision RESULTS The performance of each machine learning algorithm in classifying children with CND and children with NSCNs are provided in Table A1 below. TABLE A1. Performance metrics for three classification algorithms. Performance Metric Linear SMV Logistic Accuracy 0.93 0.91 0.95 Precision 1.00 1.00 0.97 Recall 0.84 0.77 0.92 F-measure 0.90 0.84 0.94 SVM: support vector machine; F-measure: Harmonic mean of precision and recall

Claims

1. A method of determining whether an electrically evoked compound action potential(eCAP) exists in a neural response comprising:providing a template eCAP waveform;receive a recorded neural response waveform obtained from a patient with a cochlear implant;re-sampling the recorded neural response waveform;normalizing the re-sampled neural response waveform and the template eCAP waveform by subtracting mean voltages recorded in a first time period from the template eCAP waveform and the re-sampled neural response waveform;determining a first negative (N1) peak and a trailing positive peak (P2) in each of the re-sampled neural response waveform and the template eCAP waveform;scaling the template eCAP waveform vertically (voltage) to match the N1 and P2 amplitudes from the re-sampled neural response waveform and horizontally (time) to match the N1 and P2 latencies from the re-sampled neural response waveform;trimming any portion of the re-sampled neural response waveform and the scaled template eCAP waveform to a time period where both waveforms overlap;re-sampling both the re-sampled neural response waveform and the scaled template eCAP waveform at the same time periods as each other with higher resolution sampling occurring before the first time period to place emphasis on a first part of the waveforms in a correlation analysis;calculating a correlation between the re-sampled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform; anddetermining whether the correlation between the re-sampled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform indicates that the recorded neural response waveform comprises an eCAP.

2. The method of claim 1, wherein providing the template eCAP waveform comprisesproviding an eCAP waveform that was made from a neural response measured in a child with a GJB2 genetic mutation that causes hearing loss without impacting the auditory nerve.2021273915 16 Apr 20253.     The method of any one of claims 1 or 2, wherein the cochlear implant comprises aplurality of electrodes, and the recorded neural response waveform is obtained by presenting a user-defined stimuli to one electrode of the plurality of electrodes of the cochlear implant to stimulate surrounding neurons and recording an electrical response of the surrounding electrons using a neighboring electrode in the cochlear implant.

4. The method of any one of claims 1-3, wherein re-sampling the recorded neuralresponse waveform comprises up-sampling the recorded neural response waveform via cubic spline interpolation to create a smooth re-sampled neural response waveform at a higher effective sampling rate.

5. The method of any one of claims 1-4, wherein subtracting mean voltages recorded inthe first time period from the template eCAP waveform and the re-sampled neural response waveform comprises subtracting mean voltages recorded in the first 600 g-sec from the template eCAP waveform and the re-sampled neural response waveform.

6. A system for determining whether an electrically evoked compound action potential(eCAP) exists in a neural response comprising:a memory; anda processor in communication with the memory, wherein the processor executes computer-executable instructions stored in the memory, said instructions causing the processor to:retrieve a template eCAP waveform from the memory;receive a recorded neural response waveform obtained from a patient with a cochlear implant;re-sample the recorded neural response waveform;normalize the re-sampled neural response waveform and the template eCAP waveform by subtracting mean voltages recorded in a first time period from the template eCAP waveform and the re-sampled neural response waveform;determine a first negative (N1) peak and a trailing positive peak (P2) in each of the re-sampled neural response waveform and the template eCAP waveform;scale the template eCAP waveform vertically (voltage) to match the N1 and P2 amplitudes from the re-sampled neural response waveform and horizontally (time) to match the N1 and P2 latencies from the re-sampled neural response2021273915 16 Apr 2025waveform;trim any portion of the re-sampled neural response waveform and the scaled template eCAP waveform to a time period where both waveforms overlap;re-sample both the re-sampled neural response waveform and the scaled template eCAP waveform at the same time periods as each other with higher resolution sampling occurring before the first time period to place emphasis on a first part of the waveforms in a correlation analysis;calculate a correlation between the re-sampled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform; anddetermine whether the correlation between the re-sampled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform indicates that the recorded neural response waveform comprises an eCAP.

7. The system of claim 6, wherein the template eCAP waveform comprises an eCAPwaveform that was made from a neural response measured in a child with a GJB2 genetic mutation that causes hearing loss without impacting the auditory nerve.

8. The system of any one of claims 6 or 7, wherein the cochlear implant comprises aplurality of electrodes, and the recorded neural response waveform is obtained by presenting a user-defined stimuli to one electrode of the plurality of electrodes of the cochlear implant to stimulate surrounding neurons and recording an electrical response of the surrounding neurons using a neighboring electrode of the plurality of electrodes of the cochlear implant.

9. The system of any one of claims 6-8, wherein re-sampling the recorded neuralresponse waveform comprises up-sampling the recorded neural response waveform via cubic spline interpolation to create a smooth re-sampled neural response waveform at a higher effective sampling rate.

10. The system of any one of claims 6-9, wherein subtracting mean voltages recorded in the first time period from the template eCAP waveform and the re-sampled neural response waveform comprises subtracting mean voltages recorded in the first 600 g-sec from the template eCAP waveform and the re-sampled neural response waveform.2021273915 16 Apr 2025neural response waveform and the scaled template eCAP waveform at the same time periods as each other with higher resolution sampling occurring before the first time period to place emphasis on the first part of the waveforms in the correlation analysis comprises resampling both waveforms with higher resolution sampling occurring before 600 p-sec to place emphasis on the first part of the waveforms in the correlation analysis.

12. The system of any one of claims 6-11, wherein calculating the correlation between the re-sampled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform comprises calculating a Pearson correlation between the re-sampled scaled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform.

13. The system of claim 12, wherein determining whether the correlation between the re-sampled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform indicates that the recorded neural response waveform comprises an eCAP comprises determining the Pearson correlation value is larger than 0.6 and the estimated eCAP amplitude is greater than 5 uV (noise floor of the device), an eCAP is determined to be present, otherwise, an eCAP is determined to not be present.

14. A computer-program product comprising computer-executable instructions stored on a non-transitory medium, said computer-executable instructions for performing a method of determining whether an electrically evoked compound action potential (eCAP) exists in a neural response, said method comprising:receiving a template eCAP waveform;receive a recorded neural response waveform obtained from a patient with a cochlear implant;re-sampling the recorded neural response waveform;normalizing the re-sampled neural response waveform and the template eCAP waveform by subtracting mean voltages recorded in a first time period from the template eCAP waveform and the re-sampled neural response waveform;determining a first negative (N1) peak and a trailing positive peak (P2) in each of the re-sampled neural response waveform and the template eCAP waveform;scaling the template eCAP waveform vertically (voltage) to match the N1 and P2 amplitudes from the re-sampled neural response waveform and horizontally (time) to match the N1 and P2 latencies from the re-sampled neural response waveform;2021273915 16 Apr 2025trimming any portion of the re-sampled neural response waveform and the scaled template eCAP waveform to a time period where both waveforms overlap;re-sampling both the re-sampled neural response waveform and the scaled template eCAP waveform at the same time periods as each other with higher resolution sampling occurring before the first time period to place emphasis on a first part of the waveforms in a correlation analysis;calculating a correlation between the re-sampled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform; anddetermining whether the correlation between the re-sampled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform indicates that the recorded neural response waveform comprises an eCAP.

15. The computer-program product of claim 14, wherein providing the template eCAP waveform comprises providing an eCAP waveform that was made from a neural response measured in a child with a GJB2 genetic mutation that causes hearing loss without impacting the auditory nerve.

16. The computer-program product of any one of claims 14 or 15, wherein the neural response waveform is obtained by sending a user-defined stimuli through one electrode of the cochlear implant to stimulate surrounding neurons and recording an electrical response of the surrounding electrons using a neighboring electrode in the cochlear implant.

17. The computer-program product of any one of claims 14-16, wherein re-sampling the recorded neural response waveform comprises up-sampling the recorded neural response waveform via cubic spline interpolation to create a smooth re-sampled neural response waveform at a higher effective sampling rate.

18. The computer-program product of any one of claims 14-17, wherein subtracting mean voltages recorded in the first time period from the template eCAP waveform and the re-sampled neural response waveform comprises subtracting mean voltages recorded in the first 600 ^-sec from the template eCAP waveform and the re-sampled neural response waveform.2021273915 16 Apr 2025both the re-sampled neural response waveform and the scaled template eCAP waveform at the same time periods as each other with higher resolution sampling occurring before the first time period to place emphasis on the first part of the waveforms in the correlation analysis comprises re-sampling both waveforms with higher resolution sampling occurring before 600 g-sec to place emphasis on the first part of the waveforms in the correlation analysis.

20. The computer-program product of any one of claims 14-19, wherein calculating the correlation between the re-sampled re-sampled neural response waveform and the resampled scaled template eCAP waveform comprises calculating a Pearson correlation between the re-sampled scaled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform; in particular, wherein determining whether the correlation between the re-sampled scaled re-sampled neural response waveform and the re-sampled scaled template eCAP waveform indicates that the recorded neural response waveform comprises an eCAP comprises determining the Pearson correlation value is larger than 0.6 and the estimated eCAP amplitude is greater than 5 uV (noise floor of the device), an eCAP is determined to be present, otherwise, an eCAP is determined to not be present.

Images