Browsing by Author "Emamian, Mohammad Hassan"
Now showing 1 - 7 of 7
Results Per Page
Sort Options
- Artificial Intelligence in refractive errorsPublication . Lança, Carla; Emamian, Mohammad Hassan; Grzybowski, Andrzej; Grzybowski, AndrzejUncorrected refractive errors (URE) are a leading cause of visual impairment (VI) and blindness. Big data acquisition and computer technology analysis are expanding quickly. By combining traditional techniques with artificial intelligence (AI) based tools, several studies have shown great potential in the detection, prediction, and risk stratification of URE, especially myopia. Continued improvement of imaging techniques and methods to measure refraction, coupled with the application of new AI algorithms, may be the future of clinical practice and standard of care for the diagnosis and prognosis of refractive errors. This chapter reviews how AI can be applied to the field of URE and discusses challenges to implementation into clinical practice and future directions.
- Artificial Intelligence in the diagnosis of dry eyePublication . Emamian, Mohammad Hassan; Aliyari, Roqayeh; Lança, Carla; Grzybowski, Andrzej; Grzybowski, AndrzejDry eye is a multifactorial disease, and its prevalence reaches 50% in some regions of the world. The disease is characterized by several signs and symptoms, and there are various diagnostic methods. However, diagnostic methods available have limitations, with some tests having poor reliability and reproducibility. So far, there is no gold standard test for diagnosing dry eye disease (DED), as there is no linear association between signs and symptoms, and often patients who have multiple signs of the disease report few symptoms or even have no symptoms. The above issues have led to difficulties in diagnosing DED. New imaging modalities, such as anterior segment optical coherence tomography (OCT), infrared meibography, in vivo confocal microscopy, tear interferometry, and non-invasive tear break-up time (TBUT), have emerged to allow objective measurements. However, the interpretation of results is based on subjective judgment. Artificial intelligence (AI) can help to solve these problems and contribute to decision-making by interpreting the results objectively. During the past years and especially after 2014, several research studies on the use of AI in the diagnosis, classification, and monitoring of DED have been published, and the results seem promising. Machine learning and deep learning methods have shown high sensitivity, specificity, and accuracy. Implementation of AI has significantly increased the speed of diagnosing DED and its causes; however, it has not yet been integrated into clinical practice. AI will play an important role in diagnosing and managing dry eye soon. AI can also be used for analysis of big data, which may predict estimates of DED prevalence and its risk factors. This will be a big step to identify individual risk factors for upscaling precision medicine in DED.
- Comment on: Development and validation of a novel nomogram for predicting the occurrence of myopia in schoolchildren: a prospective cohort studyPublication . Lança, Carla; Parssinen, Olavi; Mehravaran, Shiva; Nordhausen, Klaus; Emamian, Mohammad Hassan; Grzybowski, AndrzejIn the recent article published by Guo et al., the authors used data from 2nd and 3rd graders to develop a nomogram to predict myopia onset in schoolchildren. Given that myopia can progress to high myopia, which is in turn a risk factor for pathologic myopia, prediction tools are timely and relevant. We would like to provide insights into other limitations and offer suggestions that can inform future works.
- Longitudinal changes in crystalline lens thickness and power in children aged 6-12 years oldPublication . Hashemi, Hassan; Khabazkhoob, Mehdi; Azizi, Elham; Iribarren, Rafael; Lança, Carla; Grzybowski, Andrzej; Rozema, Jos J.; Emamian, Mohammad Hassan; Fotouhi, AkbarObjectives: To determine the three-year changes in crystalline lens power (LP) and thickness (LT) in children and their associated factors. Methods: Schoolchildren aged 6-12 years living in Shahroud, northeast Iran were examined in 2015 and 2018. The Bennett formula was used to calculate LP. Multiple generalized estimating equations (GEE) analysis was used for data analysis. Results: Among the 8089 examined eyes, the mean LP in Phase 1 and 2, and the three-year change were 21.61 ± 1.47D, 21.00 ± 1.42D, and -0.61 ± 0.52D, respectively. The GEE model showed that negative shifts in LP were less pronounced with increasing age (β = 0.176; p < 0.001), and were also less noticeable in hyperopes compared to emmetropes (β = 0.120; p < 0.001). The changes in LP decreased when outdoor activity increased among urban residents (β = 0.013; p = 0.039), while it increased in rural areas (β = -0.020; p = 0.047). The mean three-year change in LT was 0.002 ± 0.13 mm. Female sex and aging by one year increased the LT by 0.022 mm (P < 0.001). However, LT decreased in 6-8-year-olds, while it increased in 10-12-year-old children, both in a linear fashion. The change in LT was less in myopes than in emmetropes (β = -0.018, P-value = 0.010). Conclusion: LP decreases after three years in 6 to 12-year-old children. LT increases slightly after three years in 6 to 12-year-old children. The changes in LP and LT were associated with refractive errors, place of residence, age and gender, and outdoor activity time.
- Predicting children’s myopia risk: a Monte Carlo approach to compare the performance of machine learning modelsPublication . Artiemjew, Piotr; Cybulski, Radosław; Emamian, Mohammad Hassan; Grzybowski, Andrzej; Jankowski, Andrzej; Lança, Carla; Mehravaran, Shiva; Młyński, Marcin; Morawski, Cezary; Nordhausen, Klaus; Pärssinen, Olavi; Ropiak, KrzysztofThis study presents the initial results of the Myopia Risk Calculator (MRC) Consortium, introducing an innovative approach to predict myopia risk by using trustworthy machine-learning models. The dataset included approximately 7,945 records (eyes) from 3,989 children. We developed a myopia risk calculator and an accompanying web interface. Central to our research is the challenge of model trustworthiness, specifically evaluating the effectiveness and robustness of AI (Artificial Intelligence)/ML (Machine Learning)/NLP (Nat-ural Language Processing) models. We adopted a robust methodology combining Monte Carlo simulations with cross-validation techniques to assess model performance. Our experiments revealed that an ensemble of classifiers and regression models with Lasso regression techniques provided the best outcomes for predicting myopia risk. Future research aims to enhance model accuracy by integrating image and synthetic data, including advanced Monte Carlo simulations.
- Prevalence of anisometropia and its associated factors in school-age childrenPublication . Hashemi, Hassan; Khabazkhoob, Mehdi; Lança, Carla; Emamian, Mohammad Hassan; Fotouhi, AkbarPurpose: To determine the prevalence of anisometropia and the associated demographic and biometric risk factors in children. Methods: This cross-sectional study was conducted on the elementary school children of Shahroud, east of Iran, in 2015. All rural students were recruited, while multistage cluster sampling was used to select the students in urban areas. All children underwent optometric examinations including the measurement of uncorrected and corrected visual acuity, autorefraction, and subjective refraction with cycloplegia. Biometric components were measured using the Allegro Biograph. Myopia and hyperopia were defined as a spherical equivalent ≤-0.5 and ≥ +2.00 diopter, respectively. Students with a history of ocular trauma or lack of cycloplegic refraction at least in one eye were excluded from the study. Results: Of 6624 selected children, 5620 participated in the study. After applying the exclusion criteria, the data of 5357 students (boys: 52.8%, n = 2834) were analyzed. The mean age of the subjects was 9.2 ± 1.7 years (range: 6-12 years). The prevalence of anisometropia ≥ 1 D was 1.1% (95% CI: 0.8 to 1.4) in all children, 1.0% (95% CI: 0.7-1.3) in boys, 1.3% (95% CI: 0.8-1.7) in girls, 1.1% (95% CI: 0.8-1.4) in urban children, and 1.4% (95% CI: 0.5-2.3) in rural children. The prevalence of anisometropia was 8.8% (95% CI: 5.3-12.2) in myopic and 5.7% (95% CI: 2.8-8.5) in hyperopic children. Axial length asymmetry (OR = 40.9; 95%CI: 10.2-164.1), myopia (OR = 17.9; 95% CI: 9.4-33.9), and hyperopia (OR = 10.1; 95% CI: (5.1-19.7) were associated with anisometropia in multiple logistic regression model. More anisometropia was associated with more severe amblyopia. The odds of amblyopia (OR = 82.3: 38.2-177-3) and strabismus (OR = 17.6: 5.5-56.4) were significantly higher in anisometropic children. The prevalence of amblyopia was 21.7% in children with myopic anisometropia ≥ 3D, 66.7% in children with hyperopic anisometropia ≥ 3D, and 100% in cases with antimetropia ≥ 3D. Conclusion: The prevalence of anisometropia was low in Iranian schoolchildren. However, a high percentage of anisometropic students had amblyopia and strabismus. Axial length was the most important biometric component associated with anisometropia.
- Three-year change in refractive error and its risk factors: results from the Shahroud School Children Eye Cohort studyPublication . Lança, Carla; Emamian, Mohammad Hassan; Wong, Yee Ling; Hashemi, Hassan; Khabazkhoob, Mehdi; Grzybowski, Andrzej; Saw, Seang Mei; Fotouhi, AkbarObjectives: To determine spherical equivalent (SE) progression among children in the Shahroud School Children's Eye Cohort Study. Methods: A prospective cohort study recruited children aged 6 to 12 years in 2015 (baseline) with a follow-up in 2018. Cycloplegic autorefraction and axial length (AL) measurements were included. SE progression over 3 years was analysed in non-myopic (SE ≥ + 0.76 D), pre-myopic (PM; SE between +0.75 D and –0.49 D), low myopic (LM; SE between −0.5 D and −5.99 D), and high myopic (HM; SE ≤ − 6 D) eyes. Age, sex, near work, outdoor time, living place, parental myopia, mother’s education, and baseline SE were evaluated as risk factors for SE progression (≤ −0.50 D). Results: Data were available for 3989 children (7945 eyes). At baseline, 40.3% (n = 3205), 3.4% (n = 274) and 0.1% (n = 7) eyes had PM, LM and HM, respectively. At the 3-year follow-up, 40.5% (n = 3216), 7.5% (n = 599) and 0.2% (n = 15) eyes had PM, LM, and HM, respectively. SE progression in eyes with LM and HM was −1.08 ± 0.76 D and −1.60 ± 1.19 D, respectively. SE progression was associated with age at baseline (Odds Ratio [OR] = 1.14; 95% confidence interval [CI], 1.08–1.21), female sex (OR = 1.80; 95% CI: 1.48–2.18), near work (OR = 1.08; 95% CI: 1.02–1.14), parental myopia (OR = 1.20; 95% CI: 1.01–1.42) and baseline SE (OR = 2.28; 95% CI: 1.88–2.78). Conclusion: A myopic shift was associated with older age, female sex, near work, parental myopia, and greater myopic baseline SE. These results help identify children at risk of progression that may benefit from treatment and lifestyle counseling.