Performance and efficiency of machine learning algorithms for analyzing rectangular biomedical data

Deng, Fei; Huang, Jibing; Yuan, Xiaoling; Cheng, Chao; Zhang, Lanjing

doi:10.1038/s41374-020-00525-x

Cited by 34 publications

(19 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Indeed, HER2 has since become a critical prognostic factor and treatment target of breast cancer. 12,31,40,41 We also showed similar, or identical in some cases, five-year net survivals of nonscreened and screened invasive breast cancers, that may indirectly support and promote reduction of breast cancer screening in some populations. However, additional studies are required to evaluate the risk and benefits of reducing breast cancer screening in some patients.…”

Section: Trends In the Age-standardized Five-year Net-survival Of Scr...supporting

confidence: 56%

Changing Trends in the Proportional Incidence and Five-year Net Survival of Screened and Non-screened Breast Cancers among Women During 1995–2011 in England

Wu¹,

Wong²,

Lu³

et al. 2022

J Clin Transl Pathol

Self Cite

View full text Add to dashboard Cite

Background and objectives: Uptake of breast cancer screening has been decreasing in England since 2007. However, the associated factors are unclear. On the other hand, survival among breast cancer patients have recently increased. We conducted a quasi-experimental analysis to test whether the trend-change in proportional incidence of nonscreened cancers coincided with that in five-year net-survival. Methods: We extracted population-based proportional incidence and age-standardized five-year net-survival data from Public Health England that included English women with invasive breast cancer diagnosed during 1995-2011 (linked to death certificates, followed through 2016). Piece-wise loglinear models with change-point/joinpoint were used to estimate temporal trends. Results: Among 254,063 women in England with invasive breast cancer diagnosed during 1995-2011, there was downward-to-upward trend-change in proportional incidence of non-screened breast cancers (annual percent change [APC]=5.6 after 2007 versus APC=−3.5 before 2007, p<0.001) in diagnosis-year 2007, when a steeper upward-trend in age-standardized five-year net survival started (APC=5.7 after 2007/2008 versus APC=0.3 before 2007/2008. Net-survival difference of screened versus non-screened cancers also significantly narrowed (18% in 2007/2008 versus 5% in 2011). Similar associations were found in all strata of race, cancer stage, grade, and histology, except in Black patients or patients with stage I, stage III, or grade I cancer. Conclusions: There was a downward-to-upward trend-change in proportional incidence of non-screened breast cancers in 2007 that coincided with a steeper upward-trend in age-standardized five-year net survival among English women in 2007. Survival benefits of breast cancer screening decreased during 2007-2011. The data support reduction of breast cancer screening in some patients, but future validation studies are warranted.

show abstract

Section: Trends In the Age-standardized Five-year Net-survival Of Scr...supporting

confidence: 56%

Changing Trends in the Proportional Incidence and Five-year Net Survival of Screened and Non-screened Breast Cancers among Women During 1995–2011 in England

Wu¹,

Wong²,

Lu³

et al. 2022

J Clin Transl Pathol

Self Cite

View full text Add to dashboard Cite

show abstract

“…In addition to this study, previous studies have performed comparative analyses of various ML algorithms in the medical domain [12,13,47]. However, only few studies have considered the inference efficiency in addition to the predictive performance.…”

Section: Principal Findings and Related Workmentioning

confidence: 99%

“…Zhang et al [13] compared the simplicity of seven algorithms by assessing their memory usage and training time for 12 public biomedical data sets. In another study, Deng et al [47] assessed the inference time of decision tree, SVM, RF, and NN algorithms. In this study, we executed our efficiency evaluation by directly exploring and comparing the power consumption levels of ML algorithms.…”

Section: Principal Findings and Related Workmentioning

confidence: 99%

Energy Efficiency of Inference Algorithms for Clinical Laboratory Data Sets: Green Artificial Intelligence Study

Yu¹,

Chen²,

Huang³

et al. 2022

J Med Internet Res

View full text Add to dashboard Cite

Background The use of artificial intelligence (AI) in the medical domain has attracted considerable research interest. Inference applications in the medical domain require energy-efficient AI models. In contrast to other types of data in visual AI, data from medical laboratories usually comprise features with strong signals. Numerous energy optimization techniques have been developed to relieve the burden on the hardware required to deploy a complex learning model. However, the energy efficiency levels of different AI models used for medical applications have not been studied. Objective The aim of this study was to explore and compare the energy efficiency levels of commonly used machine learning algorithms—logistic regression (LR), k-nearest neighbor, support vector machine, random forest (RF), and extreme gradient boosting (XGB) algorithms, as well as four different variants of neural network (NN) algorithms—when applied to clinical laboratory datasets. Methods We applied the aforementioned algorithms to two distinct clinical laboratory data sets: a mass spectrometry data set regarding Staphylococcus aureus for predicting methicillin resistance (3338 cases; 268 features) and a urinalysis data set for predicting Trichomonas vaginalis infection (839,164 cases; 9 features). We compared the performance of the nine inference algorithms in terms of accuracy, area under the receiver operating characteristic curve (AUROC), time consumption, and power consumption. The time and power consumption levels were determined using performance counter data from Intel Power Gadget 3.5. Results The experimental results indicated that the RF and XGB algorithms achieved the two highest AUROC values for both data sets (84.7% and 83.9%, respectively, for the mass spectrometry data set; 91.1% and 91.4%, respectively, for the urinalysis data set). The XGB and LR algorithms exhibited the shortest inference time for both data sets (0.47 milliseconds for both in the mass spectrometry data set; 0.39 and 0.47 milliseconds, respectively, for the urinalysis data set). Compared with the RF algorithm, the XGB and LR algorithms exhibited a 45% and 53%-60% reduction in inference time for the mass spectrometry and urinalysis data sets, respectively. In terms of energy efficiency, the XGB algorithm exhibited the lowest power consumption for the mass spectrometry data set (9.42 Watts) and the LR algorithm exhibited the lowest power consumption for the urinalysis data set (9.98 Watts). Compared with a five-hidden-layer NN, the XGB and LR algorithms achieved 16%-24% and 9%-13% lower power consumption levels for the mass spectrometry and urinalysis data sets, respectively. In all experiments, the XGB algorithm exhibited the best performance in terms of accuracy, run time, and energy efficiency. Conclusions The XGB algorithm achieved balanced performance levels in terms of AUROC, run time, and energy efficiency for the two clinical laboratory data sets. Considering the energy constraints in real-world scenarios, the XGB algorithm is ideal for medical AI applications.

show abstract

“…It plays an important role in clinical diagnosis, precision treatment and health monitoring of diseases. Machine learning algorithms have been used to analyze the results of biomedical datasets, including random forests (RF), decision tree (DT), and support vector machine (SVM) [ 10 ]. Based on machine learning, clinical influencing factors of AMI were studied [ 11 ], the 1-year mortality of AMI was predicted [ 12 ], and a prediction model of arrhythmia after AMI was established [ 13 ].…”

Section: Introductionmentioning

confidence: 99%

Uncovering potential diagnostic biomarkers of acute myocardial infarction based on machine learning and analyzing its relationship with immune cells

Kang

Zhao

Jiang

et al. 2023

BMC Cardiovasc Disord

View full text Add to dashboard Cite

Background Acute myocardial infarction (AMI) is a common cardiovascular disease. This study aimed to mine biomarkers associated with AMI to aid in clinical diagnosis and management. Methods All mRNA and miRNA data were downloaded from public database. Differentially expressed mRNAs (DEmRNAs) and differentially expressed miRNAs (DEmiRNAs) were identified using the metaMA and limma packages, respectively. Functional analysis of the DEmRNAs was performed. In order to explore the relationship between miRNA and mRNA, we construct miRNA-mRNA negative regulatory network. Potential biomarkers were identified based on machine learning. Subsequently, ROC and immune correlation analysis were performed on the identified key DEmRNA biomarkers. Results According to the false discovery rate < 0.05, 92 DEmRNAs and 272 DEmiRNAs were identified. GSEA analysis found that kegg_peroxisome was up-regulated in AMI and kegg_steroid_hormone_biosynthesis was down-regulated in AMI compared to normal controls. 5 key DEmRNA biomarkers were identified based on machine learning, and classification diagnostic models were constructed. The random forests (RF) model has the highest accuracy. This indicates that RF model has high diagnostic value and may contribute to the early diagnosis of AMI. ROC analysis found that the area under curve of 5 key DEmRNA biomarkers were all greater than 0.7. Pearson correlation analysis showed that 5 key DEmRNA biomarkers were correlated with most of the differential infiltrating immune cells. Conclusion The identification of new molecular biomarkers provides potential research directions for exploring the molecular mechanism of AMI. Furthermore, it is important to explore new diagnostic genetic biomarkers for the diagnosis and treatment of AMI.

show abstract

Performance and efficiency of machine learning algorithms for analyzing rectangular biomedical data

Cited by 34 publications

References 33 publications

Changing Trends in the Proportional Incidence and Five-year Net Survival of Screened and Non-screened Breast Cancers among Women During 1995–2011 in England

Changing Trends in the Proportional Incidence and Five-year Net Survival of Screened and Non-screened Breast Cancers among Women During 1995–2011 in England

Energy Efficiency of Inference Algorithms for Clinical Laboratory Data Sets: Green Artificial Intelligence Study

Uncovering potential diagnostic biomarkers of acute myocardial infarction based on machine learning and analyzing its relationship with immune cells

Contact Info

Product

Resources

About