Accelerating Hyperparameter Tuning in Machine Learning for Alzheimer's Disease With High Performance Computing

Prediction of medication-related osteonecrosis of the jaws using machine learning methods from estrogen receptor 1 polymorphisms and clinical information

Objective

The purpose of this study was to evaluate the effect of estrogen receptor 1 (ESR1) polymorphisms on the development of medication-related osteonecrosis of the jaws (MRONJ) in women with osteoporosis.

Methods

A total of 125 patients taking bisphosphonates was evaluated the relationship between MRONJ occurrence and single nucleotide polymorphisms (SNPs) of ESR1. Clinical information was collected, including current age, treatment duration, and comorbidity. Univariate and Multivariable regression analyzes were performed to evaluate the independent predictive factors for MRONJ occurrence. Predictive models were constructed using machine learning methods such as Lasso regression, Random forest (RF), and Support vector machine (SVM). The area under the receiver-operating curve (AUROC) was used to evaluate the performance of a binary classifier.

Result

Two SNPs of ESR1 (rs4870056 and rs78177662) were significantly associated with MRONJ development. Patients with variant allele (A) of rs4870056 showed 2.45 times (95% CI, 1.03-5.87) the odds of MRONJ occurrence compared to those with wild-type homozygote (GG) after adjusting covariates. Additionally, carriers with variant allele (T) of rs78177662 had higher odds than those with wild-type homozygote (CC) (adjusted odds ratio (aOR), 2.64, 95% CI, 1.00-6.94). Among demographic variables, age ≥ 72 years (aOR, 3.98, 95% CI, 1.60-9.87) and bisphosphonate exposure ≥48 months (aOR, 3.16, 95% CI, 1.26-7.93) were also significant risk factors for MRONJ occurrence. AUROC values of machine learning methods ranged between 0.756-0.806 in the study.

Conclusion

Our study showed that the MRONJ occurrence was associated with ESR1 polymorphisms in osteoporotic women.

Hyperparameter Tuning with High Performance Computing Machine Learning for Imbalanced Alzheimer’s Disease Data

Accurate detection is still a challenge in machine learning (ML) for Alzheimer’s disease (AD). Class imbalance in imbalanced AD data is another big challenge for machine-learning algorithms working under the assumption that the data are evenly distributed within classes. Here, we present a hyperparameter tuning workflow with high-performance computing (HPC) for imbalanced data related to prevalent mild cognitive impairment (MCI) and AD in the Health and Aging Brain Study-Health Disparities (HABS-HD) project. We applied a single-node multicore parallel mode to hyperparameter tuning of gamma, cost, and class weight using a support vector machine (SVM) model with 10 times repeated fivefold cross-validation. We executed the hyperparameter tuning workflow with R’s bigmemory, foreach, and doParallel packages on Texas Advanced Computing Center (TACC)’s Lonestar6 system. The computational time was dramatically reduced by up to 98.2% for the high-performance SVM hyperparameter tuning model, and the performance of cross-validation was also improved (the positive predictive value and the negative predictive value at base rate 12% were, respectively, 16.42% and 92.72%). Our results show that a single-node multicore parallel structure and high-performance SVM hyperparameter tuning model can deliver efficient and fast computation and achieve outstanding agility, simplicity, and productivity for imbalanced data in AD applications.