Sergey Klimov, Msc, Phd Email & Phone Number

Irving, Texas, Us

Irving, Texas, Us

Irving, Texas, Us

Atlanta, Ga, Us

● Derived multiple assays, from immunohistochemistry and clino-pathological variables, to fit into a logistic regression which predicted >6x higher risk for specific site of breast metastasis over a 15-year follow-up. ● Designed a state of art breast cancer image analysis pipeline which annotated surgical tissue before using a handcrafted list of over 3k textural and spatial features and a machine learning zoo to identify patients who are at higher risk of recurrence (85% accuracy on an independent validation dataset). ● Developed a nested CNN image analysis model to predict Pancreatic Neuroendocrine tumor (PanNET) metastasis risk. The novel 3 step approach (annotation, whole slide feature extraction, outcome prediction) leveraged over 14 million parameters and provided prognostic value which outperformed any clinical variable for a cancer with very few biomarkers.● Biostatistical lead in expansive international collaborations to identify various protein expression prognostic value, depending on race, through survival analysis.● Designed a machine learning approach for determining tumor growth linking a unique in-vivo dataset and patient outcomes. Patients with fast growing tumors had almost a 4x higher chance of dying even when controlling for common clinical variables.● Patented a non-invasive approach to identifying liver fibrosis through analyzing time series ultrasound waves with varied machine learning algorithms and preprocessing.

New York, Ny, Us

● Developed, and productionalized, an AutoML model to predict members at high risk of avoidable hospitalizations due to sepsis. Using a wide collection of disparate features, including claims, procedures, behavioral, and demographics this model captured 7x more than chance to facilitate efficient, un-biased, outreach.● Created the first productionalized python model for Healthfirst, with c-level support thanks to highly intuitive visualizations and communication, which identified expecting mothers, at the earliest point, who were at high risk for a newborn with complication (low birthweight, preterm, and/or ventilation support). Directly working with the care management team to facilitate outreach towards an estimated 40 pregnant women a week.

Worldwide, Oo

● Developed, and optimized, multiple models (random forests, XGBoost, etc.) combining 25 million observations and proprietary marketing, behavioral, and clinical features to predict, on both the individual (using SAS) and county level (Python), high risk for upcoming drug abuse. Final developed model integrated a substantial data preprocessing pipeline and hyperparameter optimization to outperform longitudinal hospital care data (r2 >0.90).● Implemented end-to-end deep learning, including RNN’s and transformers, and traditional ML NLP pipelines, such as tf-idf with naïve bayes, to identify EMR based prognostic patterns which could differentiate COVID diagnosis from influenza with >80% accuracy.● Investigated temporal emergency department visits, laboratory results, and mobility data to predict both temporal and spatial features responsible for COVID outbreaks as part of the CDC surveillance effort. Used multiple algorithms (i.e. permutation, bootstrapping, LIME, and SHAP) for feature selection to identify set of most significant, and interpretable, data elements predictive for upcoming hotspots.● Investigated a cloud based de-identification pipeline and developed helper functions to synthetically create data, score outputs, and analyze performances.● Guided peer-reviewed publications, white papers, and conference efforts with commercial and academic partners.

Los Angeles, Ca, Us

● Led analysis linking single-cell morphology and genetics to predict pancreatic cancer metastasis risk.● Consulted on statistical and survival analysis for publications

Ti:Ger Certificate, Business Administration And Management, General

Doctor Of Philosophy - Phd, Biology/Mathematics And Statistics

Master Of Science (M.S.), Biology, Bioinformatics

Bachelor'S Degree, Biomedical/Medical Engineering