Stage 1.X — Meta-Design with Machine Learning is Coming, and That's a Good Thing

Background: Patients suffering from subarachnoid hemorrhage (SAH) have poor long-term outcomes. There are predictive models for ischemic and hemorrhagic stroke. However, there is paucity of models for SAH. Machine learning concepts were applied to build multi-stage Neural Networks (NN), Support Vector Machines (SVM) and Keras/Tensor Flow models to predict SAH outcomes. Methods: A database of ~800 aneurysmal SAH patients from Kasturba Medical College was utilized. Baseline variables of World Federation of Neurosurgeons 5-point scale (WFNS 1-5), age, gender, and presence/absence of hypertension and diabetes were considered in Stage 1. Stage 2 included all Stage 1 variables along with presence/absence of radiologic signs vasospasm and ischemia. Stage 3 includes earlier 2 stages and discharge Glasgow Outcome Scale (GOS 1-5). GOS at 3 months was predicted using 2-layer NN/SVM/Keras-TensorFlow models on the five point categorical scale as well as dichotomized to dead/alive and favorable (GOS 4-5) or unfavorable (GOS 1-3). Prediction accuracy of models was compared to the recorded GOS. Results: Prediction accuracy shown as percentages (See Table) for all three stages was similar for SVM, NN and Keras/TensorFlow models. Accuracy was remarkably higher with dichotomization compared to the complete five point GOS categorical scale. Conclusions: SVM, NN, and Keras-TensorFlow based machine learning models can be used to predict SAH outcomes to a high degree of accuracy. These powerful predictive models can be used to prognosticate and select patients into trials.

Download Full-text

Modeling of prognostication and differential genomic expression in the tumor microenvironment of clear cell renal cell carcinoma.

Journal of Clinical Oncology ◽

10.1200/jco.2019.37.7_suppl.557 ◽

2019 ◽

Vol 37 (7_suppl) ◽

pp. 557-557

Author(s):

Sunyoung S. Lee ◽

Seok Joon Kwon ◽

Dong Eun Lee ◽

Seongwon Lee ◽

Andrew Baird ◽

...

Keyword(s):

Machine Learning ◽

Tumor Microenvironment ◽

Clear Cell ◽

Response To Therapy ◽

Cell Renal Cell Carcinoma ◽

Genomic Expression ◽

Expression Of Genes ◽

Kaplan Meier ◽

Stage 1 ◽

Worse Prognosis

557 Background: Stroma in the tumor microenvironment (TME) influences prognosis and response to therapy. Few mathematical models exist to prognosticate patients (pts), based on mRNA expressivity in the TME. Methods: Clinical outcomes data and mRNA-seq of 533 pts with clear cell renal cancer were obtained from TCGA. Expressivity of 191 genes enriched in cellular and structural elements of TME and clinical data were analyzed via machine learning, multivariate nonlinear regression with confined optimization, and Kaplan-Meier (KM) analysis. Results: Prognostication was modeled with higher risk score (RS) representing worse prognosis in each stage (Table). P/G is the ratio of genes associated with poor (61 genes) to good (14) prognosis (refer to presentation). Based on RS, pts in each stage were clustered into 2 groups (high and low RS), showing 2 KM curves with p < 0.001 in each stage. Analysis of immune profiles in these 2 groups shows that in stage 1, expression of genes related to immune activation (IA) is not statistically different in high and low RS groups, but expression of genes related to immune inhibition (II) is higher in high RS group. In high RS groups of stage 2-4, IA genes are highly co-expressed with II genes. In high RS groups of all stages, expression of both IA and II genes increases as stage increases. In low RS groups, IA genes increase as stage increases, but II genes do not. Conclusions: Machine learning and mathematical modeling of RS and gene analysis show that IA genes are suppressed by high degree of II in high RS groups of advanced stages, contributing to worse prognosis. RS enables prognostication of pts encountered in the clinic, given genomic profiles. [Table: see text]

Download Full-text

Machine Learning for the Prediction of Progression in Patients with Acute Kidney Injury in Critical Care

10.21203/rs.3.rs-412422/v1 ◽

2021 ◽

Author(s):

Lifan Zhang ◽

Canzheng Wei ◽

Yunxia Feng ◽

Aijia Ma ◽

Yan Kang

Keyword(s):

Machine Learning ◽

Acute Kidney Injury ◽

Logistic Regression ◽

Critical Care ◽

Intensive Care ◽

Logistic Regression Model ◽

Kidney Injury ◽

Gradient Boosting ◽

Extreme Gradient Boosting ◽

Stage 1

Abstract Background: Acute kidney injury (AKI) is a serve and harmful syndrome in the intensive care unit. The purpose of this study is to develop a prediction model that predict whether patients with AKI stage 1/2 will progress to AKI stage 3. Methods: Patients with AKI stage 1/2, when they were ﬁrst diagnosed with AKI in the Medical Information Mart for Intensive Care (MIMIC-III), were included. We excluded patients who had underwent RRT or progressed to AKI stage 3 within 72 hours of the ﬁrst AKI diagnosis. We also excluded patients with chronic kidney disease (CKD). We used the Logistic regression and machine learning extreme gradient boosting (XGBoost) to build two models which can predict patients who will progress to AKI stage 3. Established models were evaluated by cross-validation, receiver operating characteristic curve (ROC), and precision-recall curves (PRC). Results: We included 25711 patients, of whom 2130 (8.3%) progressed to AKI stage 3. Creatinine, multiple organ failure syndromes (MODS), blood urea nitrogen (BUN), sepsis, and respiratory failure were the most important in AKI progression prediction. The XGBoost model has a better performance than the Logistic regression model on predicting AKI stage 3 progression (AU-ROC, 0.926; 95%CI, 0.917 to 0.931 vs. 0.784; 95%CI, 0.771 to 0.796, respectively). Conclusions: The XGboost model can better identify patients with AKI progression than Logistic regression model. Machine learning techniques may improve predictive modeling in medical research. Keywords: Acute kidney injury; Critical care; Logistic Models; Extreme gradient boosting

Download Full-text

NODE-RED Microservice Bedside System for Outcome Prediction of Patients with Suspected SEPSIS: Useful Study for the Coronavirus Outbreak and Beyond (Stage 1 of the Work)

10.36227/techrxiv.12146286 ◽

2020 ◽

Author(s):

Sabah Mohammed ◽

Jinan Fiaidhi ◽

Sami Mohammed

Keyword(s):

Machine Learning ◽

Outcome Prediction ◽

Important Research ◽

Suspected Sepsis ◽

Ongoing Research ◽

Early Stages ◽

Stage 1

This is a stage 1 of our ongoing research to build bedside machine learning predictors to identify septic shocks at early stages. It is an important research as the current COVID-19 causes such septic shocks and predicting them early save lives.

Download Full-text

Geographically Weighted Machine Learning and Downscaling for High-Resolution Spatiotemporal Estimations of Wind Speed

Remote Sensing ◽

10.3390/rs11111378 ◽

2019 ◽

Vol 11 (11) ◽

pp. 1378 ◽

Cited By ~ 4

Author(s):

Lianfa Li

Keyword(s):

Machine Learning ◽

High Resolution ◽

Wind Speed ◽

Mainland China ◽

Reanalysis Data ◽

High Temporal Resolution ◽

Residual Network ◽

Learning Stage ◽

Ensemble Machine Learning ◽

Stage 1

High-resolution spatiotemporal wind speed mapping is useful for atmospheric environmental monitoring, air quality evaluation and wind power siting. Although modern reanalysis techniques can obtain reliable interpolated surfaces of meteorology at a high temporal resolution, their spatial resolutions are coarse. Local variability of wind speed is difficult to capture due to its volatility. Here, a two-stage approach was developed for robust spatiotemporal estimations of wind speed at a high resolution. The proposed approach consists of geographically weighted ensemble machine learning (Stage 1) and downscaling based on meteorological reanalysis data (Stage 2). The geographically weighted machine learning method is based on three base learners, which are an autoencoder-based deep residual network, XGBoost and random forest, and it incorporates spatial autocorrelation and heterogeneity to boost the ensemble predictions. With reanalysis data, downscaling was introduced in Stage 2 to reduce bias and spatial abrupt (non-natural) variation in the predictions inferred from Stage 1. The autoencoder-based residual network was used in Stage 2 to adjust the difference between the averages of the fine-resolution predicted values and the coarse-resolution reanalysis data to ensure consistency. Using mainland China as a case study, the geographically weighted regression (GWR) ensemble predictions were shown to perform better than individual learners’ predictions (with an approximately 12–16% improvement in R2 and a decrease of 0.14–0.19 m/s in root mean square error). Downscaling further improved the predictions by reducing inconsistency and obtaining better spatial variation (smoothing). The proposed approach can also be applied for the high-resolution spatiotemporal estimation of other meteorological parameters or surface variables involving remote sensing images (i.e. reliable coarsely resolved data), ground monitoring data and other relevant factors.

Download Full-text

Towards Comparing Machine Learning Models to Foresee the Stages for heart disease

10.5121/csit.2021.110304 ◽

2021 ◽

Author(s):

Khalid Amen ◽

Mohamed Zohdy ◽

Mohammed Mahmoud

Keyword(s):

Machine Learning ◽

Heart Disease ◽

Early Stage ◽

Disease Stage ◽

Learning Approaches ◽

Disease Rates ◽

Stage 1 ◽

Severe Heart Disease ◽

F Measure ◽

Machine Learning Models

With the increase in heart disease rates at advanced ages, we need to put a high quality algorithm in place to be able to predict the presence of heart disease at an early stage and thus, prevent it. Previous Machine Learning approaches were used to predict whether patients have heart disease. The purpose of this work is to compare two more algorithms (NB, KNN) to our previous work [1] to predict the five stages of heart disease starting from no disease, stage 1, stage 2, stage 3 and advanced condition, or severe heart disease. We found that the LR algorithm performs better compared to the other two algorithms. The experiment results show that LR performs the best with an accuracy of 82%, followed by NB with an accuracy of 79% when all three classifiers are compared and evaluated for performance based on accuracy, precision, recall and F measure.

Download Full-text

NODE-RED Microservice Bedside System for Outcome Prediction of Patients with Suspected SEPSIS: Useful Study for the Coronavirus Outbreak and Beyond (Stage 1 of the Work)

10.36227/techrxiv.12146286.v1 ◽

2020 ◽

Author(s):

Sabah Mohammed ◽

Jinan Fiaidhi ◽

Sami Mohammed

Keyword(s):

Machine Learning ◽

Outcome Prediction ◽

Important Research ◽

Suspected Sepsis ◽

Ongoing Research ◽

Early Stages ◽

Stage 1

This is a stage 1 of our ongoing research to build bedside machine learning predictors to identify septic shocks at early stages. It is an important research as the current COVID-19 causes such septic shocks and predicting them early save lives.

Download Full-text

Machine Learning Classifiers Evaluation for Automatic Karyogram Generation from G-Banded Metaphase Images

Applied Sciences ◽

10.3390/app10082758 ◽

2020 ◽

Vol 10 (8) ◽

pp. 2758

Author(s):

Yahir Hernández-Mier ◽

Marco Aurelio Nuño-Maganda ◽

Said Polanco-Martagón ◽

María del Refugio García-Chávez

Keyword(s):

Machine Learning ◽

Classification Scheme ◽

Phase 1 ◽

Staining Technique ◽

Desktop Application ◽

Stage Classification ◽

Multiclass Classifier ◽

Stage 1 ◽

Selection Of

This work proposes the evaluation of a set of algorithms of machine learning and the selection of the most appropriate one for the classification of segmented chromosomes images acquired using the Giemsa staining technique (G-banding). The evaluation and selection of the best classification algorithms was carried out over a dataset of 119 Q-banding chromosomes images, and the obtained results were then applied to a dataset of 24 G-band chromosomes images, manually classified by an expert of the Laboratory of Cytogenetic of the Children’s Hospital of Tamaulipas. The results of evaluation of 51 classifiers yielded that the best classification accuracy for the selected features was obtained by a backpropagation neural network. One of the main contributions of this study is the proposal of a two-stage classification scheme based on the best classifier found by the initial evaluation. In stage 1, chromosome images are classified into three major groups. In stage 2, the output of phase 1 is used as the input of a multiclass classifier. Using this scheme, 82% of the IGB bank samples and 88% of the samples of a bank of images obtained with a Q-band available in the literature consisting of 119 chromosome studies were successfully classified. The proposed work is a part of an desktop application that allows cytogeneticist to automatically generate cytogenetic reports.

Download Full-text

Predicting Romantic Interest during Early Relationship Development: A Preregistered Investigation using Machine Learning

10.31219/osf.io/sh7ja ◽

2021 ◽

Author(s):

Paul Wolfe Eastwick ◽

Samantha Joel ◽

Daniel C. Molden ◽

Eli Finkel ◽

Kathleen L. Carswell

Keyword(s):

Machine Learning ◽

Theoretical Perspective ◽

Romantic Partners ◽

Relationship Development ◽

Romantic Attraction ◽

Separation Distress ◽

Data Points ◽

Early Relationship ◽

Stage 1 ◽

Romantic Interest

There are massive literatures on initial romantic attraction and established, “official” relationships. But there is a gap in our knowledge about early relationship development: the interstitial stretch of time in which people experience rising and falling romantic interest for partners who have the potential to—but often do not—become sexual or dating partners. In the current study, 208 single participants reported on 1,065 potential romantic partners across 7,179 data points over seven months. In stage 1 of the analyses, we used machine learning (specifically, Random Forests) to extract estimates of the extent to which different classes of predictors (e.g., individual differences vs. target-specific constructs) accounted for participants’ romantic interest in these potential partners (12% vs. 36%, respectively). Also, the machine learning analyses offered little support for perceiver × target moderation accounts of compatibility: the meta-theoretical perspective that some types of perceivers are likely to experience greater romantic interest for some types of targets. In stage 2, we used traditional multilevel-modeling approaches to depict growth-curve analyses for each predictor retained by the machine learning models; robust (positive) main effects emerged for many variables, including sociosexuality, gender, the potential partner’s positive attributes (e.g., attractive, exciting), attachment features (e.g., proximity seeking, separation distress), and perceived interest. We also directly tested (and found no support for) ideal partner preference-matching effects on romantic interest, which is one popular perceiver × target moderation account of compatibility. We close by discussing the need for new models and perspectives to explain how people assess romantic compatibility.

Download Full-text

Risk score and prognostication modeling based on mRNA expressivity in the tumor microenvironment of hepatocellular carcinoma.

Journal of Clinical Oncology ◽

10.1200/jco.2020.38.15_suppl.e16658 ◽

2020 ◽

Vol 38 (15_suppl) ◽

pp. e16658-e16658

Author(s):

Sunyoung S. Lee ◽

Yehia I. Mohamed ◽

Sung Hwan Lee ◽

Andrew Baird ◽

Jillian Dolan ◽

...

Keyword(s):

Machine Learning ◽

Hepatocellular Carcinoma ◽

Tumor Microenvironment ◽

Clinical Data ◽

Risk Score ◽

Cox Model ◽

Response To Therapy ◽

Age At Diagnosis ◽

Stage 1 ◽

Worse Prognosis

e16658 Background: Stromal elements in the tumor microenvironment (TME) impact prognosis and response to therapy. Advances in mRNA-sequencing improved understanding of gene expressivity, but few models exist to model prognosis in association with mRNA expression. Methods: Clinical data and mRNA-seq of 256 patients (pts) with hepatocellular carcinoma (HCC) were obtained from TCGA. The expressivity of 191 genes enriched in cellular and structural components of the TME and clinical data were analyzed using machine learning, multivariable COX model, and Kaplan-Meier (KM) analysis to model risk score (RS) for prediction of prognosis. Results: Prognostication was modeled with higher risk score (RS) representing worse prognosis. Gene expression associated with poor (P) and good (G) in stage 1 and 2 HCC was identified (refer to presentation). RS (stage 1) = 5.997 - 0.589 × (Age at diagnosis−7.979E-06) - 4.818 × (P/G−0.009); RS (stage 2) = -5.704 - 0.780 × (Age at diagnosis−9.383E-06) + 7.228 × (P/G−0.004). Based on RS, pts were clustered into 2 groups in each stage - high and low RS groups, showing two KM curves with P < 0.05, HR = 3.213 (95% CI 2.212 – 4.347) in stage 1; HR = 2.733 (95% CI 2.131 – 3.426) in stage 2, confirming the validity of RS modeling. Analysis of immune profiles in high and low RS groups shows that expression of genes associated with immunosuppressive factors, desmoplastic reaction, neutrophils, and co-inhibitory factors of T-cells are higher in high RS group in both stages (p < 0.05). Conclusions: Machine learning-assisted mathematical modeling of RS and gene analysis identified TME-related genes and gene groups that are strongly associated with worse prognosis in stage 1 and 2 of HCC. RS could potentially prognosticate pts in the clinic with available genomic profiles.

Download Full-text