Search

Objectives: To predict postoperative cochlear implant performance with heterogeneous text and numerical variables using supervised machine learning techniques., Study Design: A supervised machine learning approach comprising neural networks and decision tree-based ensemble algorithms were used to predict 1-year postoperative cochlear implant performance based on retrospective data., Setting: Tertiary referral center., Patients: One thousand six hundred four adults who received one cochlear implant from 1989 to 2019. Two hundred eighty two text and numerical objective demographic, audiometric, and patient-reported outcome survey instrument variables were included., Outcome Measures: Outcomes for postoperative cochlear implant performance were discrete Hearing in Noise Test (HINT; %) performance and binned HINT performance classification ("High," "Mid," and "Low" performers). Algorithm performance was assessed using hold-out validation datasets and were compared using root mean square error (RMSE) in the units of the target variable and classification accuracy., Results: The neural network 1-year HINT prediction RMSE and classification accuracy were 0.57 and 95.4%, respectively, with only numerical variable inputs. Using both text and numerical variables, neural networks predicted postoperative HINT with a RMSE of 25.0%, and classification accuracy of 73.3%. When applied to numerical variables only, the XGBoost algorithm produced a 1-year HINT score prediction performance RMSE of 25.3%. We identified over 20 influential variables including preoperative sentence-test performance, age at surgery, as well as specific tinnitus handicap inventory (THI), Short Form 36 (SF-36), and health utilities index (HUI) question responses as the highest influencers of postoperative HINT., Conclusion: Our results suggest that supervised machine learning can predict postoperative cochlear implant performance and identify preoperative factors that significantly influence that performance. These algorithms can help improve the understanding of the diverse factors that impact functional performance from heterogeneous data sources.

Purpose: To compare higher-order aberrations (HOAs) after small incision lenticule extraction (SMILE) in patients with and without intraoperative angle kappa adjustments., Methods: This is a retrospective case series. One hundred six eyes of 106 patients who underwent SMILE at Tianjin Eye Hospital (Tianjin Medical University, Tianjin, China) for correction of myopia and myopic astigmatism were divided into 2 groups. The first group consisted of eyes with intraoperative angle kappa adjustment and the second group consisted of eyes without adjustment. Preoperative and postoperative visual outcome, refraction, and HOA measurements at 1 and 3 months were compared., Results: At the pupil size of 6 mm, vertical coma at 1 and 3 months after SMILE for the angle kappa-adjusted group was 0.153 ± 0.107 and 0.157 ± 0.094 μm, which were significantly lower than those of the nonadjusted group (0.204 ± 0.117 and 0.203 ± 0.113 μm, respectively) (P = 0.026 at 1 mo, P = 0.047 at 3 mo). The change in vertical coma between preoperative and postoperative measurements was 0.011 ± 0.136 and 0.023 ± 0.129 μm at 1 and 3 months postoperatively for the angle kappa-adjusted group, which were lower than those of the nonadjusted group (0.082 ± 0.165 and 0.085 ± 0.150 μm, respectively) (P = 0.023 at 1 mo, P = 0.045 at 3 mo). Subgroup analysis for eyes with large angle kappa demonstrated that the vertical coma was significantly less in the angle kappa-adjusted group at both 1 and 3 months (P = 0.009, P = 0.043, respectively). No significant correlation was observed between angle kappa and HOAs in the angle kappa-adjusted group., Conclusions: Adjustment of angle kappa during SMILE resulted in less HOAs. It would provide more insight on how to optimize treatment centration in SMILE.

Objective: The use of machine learning technology to automate intellectual processes and boost clinical process efficiency in medicine has exploded in the past 5 years. Machine learning excels in automating pattern recognition and in adapting learned representations to new settings. Moreover, machine learning techniques have the advantage of incorporating complexity and are free from many of the limitations of traditional deterministic approaches. Cochlear implants (CI) are a unique fit for machine learning techniques given the need for optimization of signal processing to fit complex environmental scenarios and individual patients' CI MAPping. However, there are many other opportunities where machine learning may assist in CI beyond signal processing. The objective of this review was to synthesize past applications of machine learning technologies for pediatric and adult CI and describe novel opportunities for research and development., Data Sources: The PubMed/MEDLINE, EMBASE, Scopus, and ISI Web of Knowledge databases were mined using a directed search strategy to identify the nexus between CI and artificial intelligence/machine learning literature., Study Selection: Non-English language articles, articles without an available abstract or full-text, and nonrelevant articles were manually appraised and excluded. Included articles were evaluated for specific machine learning methodologies, content, and application success., Data Synthesis: The database search identified 298 articles. Two hundred fifty-nine articles (86.9%) were excluded based on the available abstract/full-text, language, and relevance. The remaining 39 articles were included in the review analysis. There was a marked increase in year-over-year publications from 2013 to 2018. Applications of machine learning technologies involved speech/signal processing optimization (17; 43.6% of articles), automated evoked potential measurement (6; 15.4%), postoperative performance/efficacy prediction (5; 12.8%), and surgical anatomy location prediction (3; 7.7%), and 2 (5.1%) in each of robotics, electrode placement performance, and biomaterials performance., Conclusion: The relationship between CI and artificial intelligence is strengthening with a recent increase in publications reporting successful applications. Considerable effort has been directed toward augmenting signal processing and automating postoperative MAPping using machine learning algorithms. Other promising applications include augmenting CI surgery mechanics and personalized medicine approaches for boosting CI patient performance. Future opportunities include addressing scalability and the research and clinical communities' acceptance of machine learning algorithms as effective techniques.

Objectives: To evaluate recent studies on available therapies for meibomian gland dysfunction (MGD)., Methods: A literature search on recent publications, within the last five years, concerning treatment options for MGD was performed., Results: A total of 35 articles were reviewed after curation by the authors for relevance. In general, all modalities of treatments were shown to have clinical efficacy in alleviating dry eye signs and symptoms, although the extent of improvement and persistency of outcomes varied between the different treatments. Evidence from published studies demonstrate that thermal pulsation produces the longest lasting effect per treatment, but it also incurs the highest per-treatment cost. Reusable methods for warm compress with lipid/semi-fluorinated alkane-containing eye drops are recommended as first-line treatment for mild-to-moderate dry eye patients, because this option is most technically feasible and cost-effective in clinical practice. Intense pulsed light (IPL) therapy and thermal pulsation may be suitable as second line for patients unresponsive to warm compress therapy; however, their respective limitations need to be considered. For refractory MGD with features of periductal fibrosis or severe blepharitis, supplementary treatment with meibomian gland probing or oral antibiotics may be used., Conclusions: All eight forms of treatments, including self-applied eyelid warming, thermal pulsation, IPL, MG probing, antibiotics, lipid-containing eye drops, and perfluorohexyloctane, were effective against MGD, although with varying extent of clinical improvements. A better understanding on the mechanisms of actions may guide physicians to make better treatment decisions targeting the root causes.

Purpose: To compare the repeatability and agreement between a swept-source biometer and a Scheimpflug biometer in cataract patients., Methods: Three consecutive measurements were obtained using a swept-source biometer (IOLMaster 700) and a Scheimpflug biometer (AL-Scan) in 52 eyes of 52 patients. Keratometry, central corneal thickness (CCT), anterior chamber depth (ACD), axial length, and white-to-white (WTW) distance were recorded. Astigmatism values were transformed into vector components of J0 and J45. Intraoperator repeatability was analyzed using intraclass correlation coefficients (ICCs) and reproducibility coefficients (RCs). Agreement of measurements between the two devices was evaluated using the Bland-Altman method., Results: The IOLMaster 700 showed higher ICCs and lower RCs for the mean keratometry (Km) (P≤0.018), CCT (P≤0.027), and ACD (P≤0.001) measurements, whereas the AL-Scan showed higher ICC and lower RC for the J45 vector component of astigmatism at the 2.4-mm zone (P≤0.034). Both the devices had excellent repeatability (ICC=0.999) in axial length measurement. Systematic differences were found in Km, CCT, ACD, and WTW (P≤0.018) between the devices. The mean difference for Km was -0.196 and -0.144 D measured at the 2.4-mm zone and 3.3-mm zone, respectively. The corresponding mean difference for CCT, ACD, and WTW distance was 14.92 μm, -0.017 mm, and 0.283 mm, respectively. These differences led to a statistically significant but clinically insignificant difference in the prediction of intraocular lens power., Conclusions: This study showed significant differences in anterior segment measurement repeatability and agreement between a swept-source biometer and a Scheimpflug biometer in eyes with cataract.

Purpose: To evaluate the history of densitometric data in patients with keratoconus undergoing corneal cross-linking., Methods: Twenty-two eyes of 22 patients with keratoconus were included. Corneal tomography and densitometry measurements were performed before and after accelerated corneal cross-linking. The duration of corneal haze was defined as the time between cross-linking and densitometry measurements returning to the preoperative value. Survival analysis of corneal haze after cross-linking was performed. Preoperative and postoperative corneal densitometry, maximum keratometry (Kmax), and central corneal thickness were compared., Results: The duration of corneal haze was 18.2 ± 3.8 months at the first zone of 0 to 2 mm and 10.9 ± 2.5 months at the second zone of 2 to 6 mm. There was no change in Kmax between the preoperative period and the time at which corneal haze resolved (P = 0.394 at the first zone; P = 0.658 at the second zone). Compared with the measurement taken at resolution of corneal haze, Kmax at 1 year after haze resolution was lower (62.0 ± 9.9 D to 61.2 ± 9.9 D, P = 0.008 at the first zone; 63.6 ± 10.9 D to 62.5 ± 10.1 D, P = 0.016 at the second zone). There was a decrease of central corneal thickness between the preoperative period and the time at which corneal haze resolved (470.8 ± 34.1 μm to 464.8 ± 34.5 μm, P = 0.047 at the first zone; 465.0 ± 35.3 μm to 454.7 ± 37.2 μm, P = 0.001 at the second zone), but it remained unchanged after haze resolution (P = 0.146 at the first zone; P = 0.067 at the second zone)., Conclusions: Corneal cross-linking halted keratoconus progression when detectable haze was present. There was continuous corneal flattening measured at 1 year after haze resolution. Thinning of the cornea was seen only when haze was detectable after cross-linking.

Objectives: To compare the various Pentacam-measured K-readings with the clinical history method (CHM) in eyes that have undergone myopic laser in situ keratomileusis (LASIK)., Methods: In this prospective study, Pentacam examination was performed in 71 eyes 1 month after myopic LASIK. The true net power (TNP) 4 mm, total corneal refractive power (TCRP) 4 mm, equivalent K-reading (EKR) 4.0 mm, and EKR 4.5 mm obtained from the same scan were compared with the K derived from CHM., Results: The average baseline spherical equivalence was -5.44±2.38 D. After LASIK, the mean KCHM was 37.67±2.13 D, TCRP4mm was 37.14±1.79 D, TNP4mm was 36.88±1.76 D, EKR4.0mm was 37.58±1.94 D, and EKR4.5mm was 37.51±1.94 D. TCRP4mm, TNP4mm, and EKR4.5mm showed a statistically significant deviation from the KCHM, with the mean error being 0.53 D, 0.79 D, and 0.16 D, respectively (P<0.05). Only the EKR4.0mm showed no statistically significant difference from the KCHM (mean error 0.09 D, P=0.23). The EKR4.0mm also had the narrowest 95% limits of agreement (LoA) (-1.10 to +1.28 D), whereas both TCRP4mm and TNP4mm had a wider LoA (-0.88 to +1.95 D and -0.62 to +2.20 D, respectively). All four Pentacam K-readings had a strong and statistically significant correlation with the KCHM., Conclusions: Using the CHM as reference, the EKR4.0mm demonstrated the closest agreement when compared with the EKR4.5mm, TNP4mm, and TCRP4mm obtained from the same scan.

Purpose: To investigate the possible risk factors and clinical effects of black areas in small-incision lenticule extraction., Methods: Thirty-one cases and 126 controls were enrolled in a case-control study from the database of a single center from September 2012 to July 2017. A retrospective analysis of surgical procedures with video and medical records was performed. Logistic regression analysis was used to assess the possible risk factors for black areas between cases with black areas and controls without black areas., Results: In the study, 19 eyes (61.3%) with black areas were associated with the presence of surface debris, 10 of which had repeated suction during docking and centration (P < 0.001). There was a significant difference in the energy of the femtosecond laser used for the cases (133 ± 4.04 nJ) and controls (138 ± 3.26 nJ) (P < 0.001). The odds ratio for the energy of the laser was 0.180 (95% confidence interval, 0.099-0.328, P < 0.001). Eyes with black areas had poorer uncorrected vision at 1 and 7 days and 1 month postoperatively (0.06, 0.01, and 0.00 logMAR, respectively) compared with controls (-0.01, -0.07, and -0.05 logMAR, respectively) (P < 0.005). At month 3, the safety index was lower in the cases (1.14 ± 0.47) compared with controls (1.36 ± 0.28) (P = 0.003). The postoperative spherical equivalent in the cases was more myopic than in controls (P = 0.036), except for month 3 (P = 0.183)., Conclusions: Black areas in small-incision lenticule extraction could be caused by repeated suction, presence of surface debris, and a lower energy of the femtosecond laser. These factors could result in rough dissection and slower recovery of visual acuity.

Purpose: To investigate the effect of the learning curve for small-incision lenticule extraction during the first 2 years of experience., Methods: Small-incision lenticule extraction was performed using the 500-kHz VisuMax femtosecond laser (Carl Zeiss Meditec) by the same surgeon. The initial 100 patients since the surgeon started operating independently were considered as group 1; the recent 100 patients were considered as group 2. The same laser settings and technique were used. The visual and refractive outcomes were compared between groups at postoperative 1 week and 6 months. Vector analysis was performed for eyes with astigmatic correction., Results: Two hundred right eyes of 200 patients were included. Age, preoperative corrected visual acuity, manifest refraction, and central corneal thickness were similar between groups (P ≤ 0.154). Postoperatively, the efficacy index at 1 week was better in group 2 (group 1: 0.85 ± 0.16 vs. group 2: 0.91 ± 0.10, P = 0.019) but was similar between groups at 6 months (group 1: 0.91 ± 0.14 vs. group 2: 0.94 ± 0.08, P = 0.181). The safety index was higher in group 2 at 1 week (group 1: 0.93 ± 0.10 vs. group 2: 0.95 ± 0.08, P = 0.045) and 6 months postoperatively (group 1: 0.97 ± 0.07 vs. group 2: 0.99 ± 0.03, P = 0.011). Vector analysis showed that postoperative residual astigmatism and misalignment of astigmatic correction were lower in group 2 than in group 1 (P ≤ 0.039) at 1 week and 6 months. The duration of docking and that of lenticule extraction was shorter in group 2 (P ≤ 0.034)., Conclusions: Our study showed that faster visual recovery, better safety profile, and more accurate astigmatic correction could be attained with increasing surgical experience.

Purpose: To study the effect of myopia on the pattern change in higher-order aberrations after small incision lenticule extraction., Methods: Sixty eyes of 60 patients were included: low myopia (≤-3.00 D), moderate myopia (-3.00 D to -6.00 D), and high myopia (≥-6.00 D). Total higher-order aberrations (tHOA), vertical coma ((Equation is included in full-text article.)), horizontal coma ((Equation is included in full-text article.)), and spherical aberration ((Equation is included in full-text article.)) were measured preoperatively and at postoperative 3 months., Results: At the end of 3 months, tHOA changed significantly compared with the preoperative values (P < 0.05), except for (Equation is included in full-text article.)and (Equation is included in full-text article.)in the low myopia group. The change in (Equation is included in full-text article.), (Equation is included in full-text article.), and (Equation is included in full-text article.)in the moderate group (-0.299, -0.175, and 0.108 μm) was 2.020, 4.861, and 4.696 times higher than the low group (-0.148, -0.036, 0.023 μm) (P = 0.002, 0.001, 0.001), respectively. The value in the high group (-0.331, -0.192, 0.154 μm) was 1.107, 1.097 (P = 0.478, 0.665), and 1.426 times (P = 0.047) higher than the moderate group. The degree of myopia was positively correlated with Δ(Equation is included in full-text article.)(r = 0.447; P < 0.001) and Δ(Equation is included in full-text article.)(r = 0.496; P < 0.001), and negatively correlated with ΔtHOA (r = -0.363, P = 0.004) and Δ(Equation is included in full-text article.)(r = -0.599; P < 0.001)., Conclusions: The study showed a different pattern of change in ocular aberrations after small incision lenticule extraction in patients with varying degrees of myopia. In patients with low myopia, there was no increase in (Equation is included in full-text article.)or (Equation is included in full-text article.). In high myopia, however, (Equation is included in full-text article.)increased with the degree of myopia, whereas the rising rate of coma was slowing.

Purpose: To compare tear osmolarity (TO) and other dry eye parameters in rheumatoid arthritis (RA) patients with or without secondary Sjogren syndrome (sSS)., Methods: Consecutive patients with RA were divided into a sSS group and no-sSS group using conventional diagnostic criteria by rheumatologists using symptomatology, Schirmer test score, and anti-Ro or anti-La autoantibody status. The TO, Ocular Surface Disease Index, dry eye disease (DED) parameters [such as tear breakup time (TBUT) and corneal staining score] and the systemic inflammatory markers [erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP)] were compared. Correlation analyses between TO and the DED parameters and inflammatory markers were also performed., Results: A total of 42 cases with mean age 54.8 ± 12.3 were included, with 12 patients (29%) having sSS and 30 (71%) without sSS. TO was increased in both groups (329 ± 20 and 319 ± 25 mOsm/L, respectively), but no statistically significant difference was found between the 2 groups (P = 0.126). RA with sSS had significantly shorter TBUT, higher corneal staining score, and ESR CRP levels (P < 0.05). TO did not correlate with the Schirmer test score, but had significant positive correlations with age, corneal staining score, ESR, and CRP levels, and a significant negative correlation with TBUT., Conclusions: TO was increased in RA patients with or without sSS. There was no significant correlation between TO and the Schirmer test score, and the physician could not use TO to diagnose sSS. However, TO correlated well with both DED parameters (TBUT and corneal staining score) and systemic inflammatory markers (ESR and CRP).

Background: Public access defibrillation programs can improve survival after out-of-hospital cardiac arrest, but automated external defibrillators (AEDs) are rarely available for bystander use at the scene. Drones are an emerging technology that can deliver an AED to the scene of an out-of-hospital cardiac arrest for bystander use. We hypothesize that a drone network designed with the aid of a mathematical model combining both optimization and queuing can reduce the time to AED arrival., Methods: We applied our model to 53 702 out-of-hospital cardiac arrests that occurred in the 8 regions of the Toronto Regional RescuNET between January 1, 2006, and December 31, 2014. Our primary analysis quantified the drone network size required to deliver an AED 1, 2, or 3 minutes faster than historical median 911 response times for each region independently. A secondary analysis quantified the reduction in drone resources required if RescuNET was treated as a large coordinated region., Results: The region-specific analysis determined that 81 bases and 100 drones would be required to deliver an AED ahead of median 911 response times by 3 minutes. In the most urban region, the 90th percentile of the AED arrival time was reduced by 6 minutes and 43 seconds relative to historical 911 response times in the region. In the most rural region, the 90th percentile was reduced by 10 minutes and 34 seconds. A single coordinated drone network across all regions required 39.5% fewer bases and 30.0% fewer drones to achieve similar AED delivery times., Conclusions: An optimized drone network designed with the aid of a novel mathematical model can substantially reduce the AED delivery time to an out-of-hospital cardiac arrest event., (© 2017 American Heart Association, Inc.)

Swept-source optical coherence tomography (OCT) is the latest advancement in anterior segment imaging. There are limited data regarding its performance after laser in situ keratomileusis (LASIK). We compared the reliability of swept-source OCT and Scheimpflug imaging for evaluation of corneal parameters in refractive surgery candidates with myopia or myopic astigmatism. Three consecutive measurements were obtained preoperatively and 1 year postoperatively using swept-source OCT and Scheimpflug imaging. The study parameters included central corneal thickness (CCT), thinnest corneal thickness (TCT), keratometry at steep (Ks) and flat (Kf) axes, mean keratometry (Km), and, anterior and posterior best fit spheres (Ant and Post BFS). The main outcome measures included reliability of measurements before and after LASIK was evaluated using intraclass correlation coefficient (ICC) and reproducibility coefficients (RC). Association between the mean value of corneal parameters with age, spherical equivalent (SEQ), and residual bed thickness (RBT) and association of variance heterogeneity of corneal parameters and these covariates were analyzed. Twenty-six right eyes of 26 participants (mean age, 32.7 ± 6.9 yrs; mean SEQ, -6.27 ± 1.67 D) were included. Preoperatively, swept-source OCT demonstrated significantly higher ICC for Ks, CCT, TCT, and Post BFS (P ≤ 0.016), compared with Scheimpflug imaging. Swept-source OCT demonstrated significantly smaller RC values for CCT, TCT, and Post BFS (P ≤ 0.001). After LASIK, both devices had significant differences in measurements for all corneal parameters (P ≤ 0.015). Swept-source OCT demonstrated a significantly higher ICC and smaller RC for all measurements, compared with Scheimpflug imaging (P ≤ 0.001). Association of variance heterogeneity was only found in pre-LASIK Ant BFS and post-LASIK Post BFS for swept-source OCT, whereas significant association of variance heterogeneity was noted for all measurements except Ks and Km for Scheimpflug imaging.This study reported higher reliability of swept-source OCT for post-LASIK corneal measurements, as compared with Scheimpflug imaging. The reliability of corneal parameters measured with Scheimpflug imaging after LASIK was not consistent across different age, SEQ, and RBT measurements. These factors need to be considered during follow-up and evaluation of post-LASIK patients for further surgical procedures.