Mo Hassanpour

Mo Hassanpour

Long Beach, California, United States
1K followers 500+ connections

About

Harnessing over a decade of expertise in machine learning and healthcare analytics, I…

Articles by Mo

Contributions

Activity

Experience

  • Centene Corporation Graphic

    Centene Corporation

    Long Beach, California, United States

  • -

    Long Beach, California, United States

  • -

    Inglewood, California, United States

  • -

    Long Beach, California, United States

  • -

    Los Angeles, California, United States

  • -

    El Segundo, California, United States

Education

  • University of Illinois Urbana-Champaign Graphic

    University of Illinois Urbana-Champaign

    -

    Program = {
    Machine Learning: 4 courses,
    Data Visualization: 1 course,
    Data Mining: 1 course,
    Cloud Computing: 1 course
    }

    Transcript = {
    Advanced Statistical Methods in R : A,
    Applied Machine Learning (Python): B,
    Practical Statistical Learning (R): B,
    Deep Learning for Health Care (Python, pyTorch): B,
    Data Visualization (Tableau and js.D3): A
    Introduction to Data Mining: TBD,
    Cloud Computing Applications: TBD
    }

  • -

    Related Courses:

    - Multivariate Analysis (including paired comparisons of multivariate means, repeated measures, multivariate regression, principal components, discrimination and classification)
    - Survey Sampling
    - Parametric and non-parametric bootstrapping and resampling Methods
    - Time Series Analysis
    - Java Programming
    - SAS Programming
    - General Linear Models (including two-way mixed effects and nested designs, RCBD, Latin squares, and linear regression)
    -…

    Related Courses:

    - Multivariate Analysis (including paired comparisons of multivariate means, repeated measures, multivariate regression, principal components, discrimination and classification)
    - Survey Sampling
    - Parametric and non-parametric bootstrapping and resampling Methods
    - Time Series Analysis
    - Java Programming
    - SAS Programming
    - General Linear Models (including two-way mixed effects and nested designs, RCBD, Latin squares, and linear regression)
    - Generalized Linear Models (including logistic regression, and log-linear models for contingency tables)

  • -

Volunteer Experience

  • School on Wheels, Inc. Graphic

    Math tutor

    School on Wheels, Inc.

    - Present 9 years 8 months

    Education

    High school Math tutor for homeless teen.

  • United Way of Greater Los Angeles Graphic

    Campaign Volunteer

    United Way of Greater Los Angeles

    - Present 6 years

    Poverty Alleviation

    EveryoneIn is a local campaign initiative backed by United Way to eliminate homelessness in Long Beach.

  • Red Cross Blood Services Graphic

    Donor

    Red Cross Blood Services

    - Present 7 years

    Health

    115 units of platelets donated so far. These are donated with two needles in an apherisis machine for 90 to 120 minute. They are used for cancer patients primarily, but also for burn and severe trauma victims.

Projects

  • Deep Learning for Healthcare

    -

    I applied and adapted cutting-edge research on hospital readmission prediction, leveraging my proficiency in deep learning and healthcare analytics. My approach involved the innovative use of a neural network framework that melded Recurrent Neural Networks (RNNs) with latent topic models. This methodology was rigorously tested and confirmed to align with existing performance standards, establishing the CONTENT model’s robustness in predicting hospital readmissions specifically for patients with…

    I applied and adapted cutting-edge research on hospital readmission prediction, leveraging my proficiency in deep learning and healthcare analytics. My approach involved the innovative use of a neural network framework that melded Recurrent Neural Networks (RNNs) with latent topic models. This methodology was rigorously tested and confirmed to align with existing performance standards, establishing the CONTENT model’s robustness in predicting hospital readmissions specifically for patients with congestive heart failure. The model’s predictive capabilities were significantly enhanced by incorporating a deep contextual understanding of clinical concepts, enabling precise analysis of electronic health records to identify nuanced patterns and structures. Furthermore, my strategic analysis facilitated the development of targeted intervention strategies aimed at reducing readmissions, thereby improving patient care outcomes and delivering substantial savings for healthcare systems.

  • Movie Recommendation System

    -

    I spearheaded the development of a sophisticated Movie Recommendation System, designed to enhance user experience through personalized movie suggestions. The system featured a genre-based recommendation module that leveraged the MovieLens 1M Dataset, categorizing films into "Highly Rated" and "Trending" based on user ratings and current popularity. To further refine the system's accuracy, I employed User-Based Collaborative Filtering (UBCF), creating a detailed rating matrix to apply cosine…

    I spearheaded the development of a sophisticated Movie Recommendation System, designed to enhance user experience through personalized movie suggestions. The system featured a genre-based recommendation module that leveraged the MovieLens 1M Dataset, categorizing films into "Highly Rated" and "Trending" based on user ratings and current popularity. To further refine the system's accuracy, I employed User-Based Collaborative Filtering (UBCF), creating a detailed rating matrix to apply cosine similarity measures. This approach allowed for the prediction of user preferences with a high degree of accuracy by analyzing rating patterns among users. Additionally, I integrated Item-Based Collaborative Filtering (IBCF), focusing on the relationships between movies themselves. By normalizing data and calculating item similarities, this method provided a nuanced model-based recommendation strategy, significantly improving the system's capability to offer relevant and appealing movie suggestions based on individual user tastes.

  • Movie Review Sentiment Analysis

    -

    I refined a sentiment analysis model for movie reviews by incorporating Lasso logistic regression, significantly improving its accuracy and efficiency. By analyzing 50,000 IMDB reviews, I pinpointed crucial sentiment indicators, streamlining the model's vocabulary for sharper insights. Further enhancements were achieved by developing and fine-tuning a Document Term Matrix (DTM) through advanced text preprocessing and vocabulary pruning, resulting in a highly effective classification model that…

    I refined a sentiment analysis model for movie reviews by incorporating Lasso logistic regression, significantly improving its accuracy and efficiency. By analyzing 50,000 IMDB reviews, I pinpointed crucial sentiment indicators, streamlining the model's vocabulary for sharper insights. Further enhancements were achieved by developing and fine-tuning a Document Term Matrix (DTM) through advanced text preprocessing and vocabulary pruning, resulting in a highly effective classification model that excelled in predicting movie review sentiments.

  • Walmart Sales Forecasting

    -

    I leveraged machine learning and advanced analytics to predict weekly sales across different categories, significantly enhancing retail operation planning. By implementing Principal Components Regression (PCR), I streamlined data to improve forecast accuracy and efficiency. Careful optimization of data pre-processing aligned training data with forecast periods, ensuring high model accuracy across ten bi-monthly forecast periods, evidenced by an average Weighted Mean Absolute Error (WMAE) of…

    I leveraged machine learning and advanced analytics to predict weekly sales across different categories, significantly enhancing retail operation planning. By implementing Principal Components Regression (PCR), I streamlined data to improve forecast accuracy and efficiency. Careful optimization of data pre-processing aligned training data with forecast periods, ensuring high model accuracy across ten bi-monthly forecast periods, evidenced by an average Weighted Mean Absolute Error (WMAE) of 1583.40. I also devised a custom training method using historical sales data to refine future sales predictions, demonstrating the impact of precise data analysis and machine learning on retail forecasting efficiency.

  • Advanced ECG Classification with MINA: Multilevel Attention-Guided CNN+RNN Model

    -

    I spearheaded the development and implementation of the MINA model, a sophisticated integration of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), designed for the binary classification of ECG signals. This model accurately differentiates between atrial fibrillation and normal sinus rhythms. A key innovation of this project was the creation of a Knowledge-guided Attention Module that incorporates three unique attention mechanisms—focusing on beat, rhythm, and frequency…

    I spearheaded the development and implementation of the MINA model, a sophisticated integration of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), designed for the binary classification of ECG signals. This model accurately differentiates between atrial fibrillation and normal sinus rhythms. A key innovation of this project was the creation of a Knowledge-guided Attention Module that incorporates three unique attention mechanisms—focusing on beat, rhythm, and frequency levels of ECG signals—for in-depth signal analysis. This pioneering approach to multilevel attention in ECG signal interpretation marks a significant leap forward in medical signal processing, enhancing the model's diagnostic accuracy. The successful application and validation of MINA underscore its potential to revolutionize atrial fibrillation detection.

  • Deep Learning for Healthcare published research replication project.

    -

    In this project another student and I replicated a paper titled, "Readmission prediction via deep contextual embedding of clinical concepts."

    Download
    Problem: Hospital readmission following admission for congestive heart failure is harmful to patients and costly.

    Solution: The goal of this paper was to target more intense readmission prevention interventions to patients at risk for readmission by predicting readmission. This is acheived by applying the proposed deep learning…

    In this project another student and I replicated a paper titled, "Readmission prediction via deep contextual embedding of clinical concepts."

    Download
    Problem: Hospital readmission following admission for congestive heart failure is harmful to patients and costly.

    Solution: The goal of this paper was to target more intense readmission prevention interventions to patients at risk for readmission by predicting readmission. This is acheived by applying the proposed deep learning (neural network) architecture to clinical note electronic health records related to congestive heart failure. Recurrent neural network and latent topic models are combined in their model. These complement each other because RNNs are good at capturing the local structure of a word sequence, but they may have difficulty, "remembering" long-range interactions. In contrast, latent topic models don’t consider word order, but they do capture the general structure of a document.

    Result: We achieved performance metric results within the ranges the authors published while applying the CONTENT model to synthetic data.

  • Heart Failure Prediction Enhancement Using Advanced Autoencoder Architectures

    -

    I crafted an interpretable model for heart failure prediction using the RETAIN framework, which incorporates an advanced Recurrent Neural Network (RNN) with a reverse time attention mechanism. This model demonstrated high accuracy on the MIMIC-III dataset by applying dual attention mechanisms for a detailed evaluation of patient visits. Innovative embedding techniques were used to convert diagnosis codes into actionable data, enhancing the model's precision. By developing a unique context…

    I crafted an interpretable model for heart failure prediction using the RETAIN framework, which incorporates an advanced Recurrent Neural Network (RNN) with a reverse time attention mechanism. This model demonstrated high accuracy on the MIMIC-III dataset by applying dual attention mechanisms for a detailed evaluation of patient visits. Innovative embedding techniques were used to convert diagnosis codes into actionable data, enhancing the model's precision. By developing a unique context vector technique and leveraging encoder hidden states, the model's interpretability in healthcare analytics was significantly improved. This work not only proved the model's capability in processing complex datasets but also marked a step forward in interpretable machine learning applications in healthcare.

Recommendations received

View Mo’s full profile

  • See who you know in common
  • Get introduced
  • Contact Mo directly
Join to view full profile

Other similar profiles

Explore collaborative articles

We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.

Explore More

Add new skills with these courses