Classification | Ukant Jadia

ECG-Lense: Cardiac ECG Classification (IEEE Published)

Thu, 01 Aug 2024 00:00:00 +0000

Published at ETNCC 2024 · IEEE · DOI: 10.1109/ETNCC63262.2024.10767459

Headline numbers

80% classification accuracy
90% ROC-AUC, the best of all six models tested
21,837 ECG records · 18,885 patients · 12-lead · PTB-XL dataset

What it is

ECG-Lense is the deep CNN architecture I built and benchmarked against five other models for classifying cardiac ECG signals. Three classical ML, three deep learning, head-to-head on the same dataset and same metrics. ECG-Lense wins.

Model	Type
Decision Tree	Classical
Random Forest	Classical
Logistic Regression	Classical
Simple CNN	Deep
LSTM	Deep
ECG-Lense (Complex CNN)	Deep (proposed)

What I actually did

Worked off the PTB-XL dataset: clinical 12-lead ECGs, 10 seconds each, sampled at 100 Hz, with hierarchical diagnostic labels (5 superclasses, 24 subclasses)
Preprocessed signals with wfdb and standard ML libraries
Implemented a stationary wavelet transform (SWT) for data augmentation. The dataset is imbalanced and the augmentation increased sample diversity while preserving signal characteristics
Trained the deep models on raw ECG signals so feature extraction would be automatic
Evaluated on accuracy, precision, recall, F1, and ROC-AUC

What broke and what I learned

Class imbalance was the actual problem. First CNN runs hit a wall around 70% accuracy because rare cardiac classes were under-represented. SWT-based augmentation unlocked the next 10 points.
Raw signals beat hand-crafted features. Classical ML on extracted features stalled. The deep models, given raw signals, found patterns the feature engineering had thrown away.
More layers wasn’t the answer. ECG-Lense is “complex” relative to the simple CNN baseline, but each block earns its place. Depth for depth’s sake didn’t help.

Six months of work. A+ thesis grade. Submitted to IEEE during the project, accepted, presented and published at ETNCC 2024.

Stack

Python · TensorFlow · Keras · scikit-learn · MLFlow · wfdb · NumPy · pandas

Acknowledgements

Mentor: Dr. Kamal Kant Hiran (PhD). He pushed me to write the paper, guided the work end-to-end. Supervisors: Alok Kumar, Dr. Chandani Joshi, Harsh Bansal.

Dataset images

The dataset comprises 21,837 clinical 12-lead ECG records of 10 seconds in length from 18,885 patients. Hierarchical diagnostic organization: 5 superclasses, 24 sub-classes.
Metadata in ptbxl_database.csv: each ECG identified by a unique ecg_id with SCP codes for multi-label classification.

Demographic overview of patients in PTB-XL.

The dataset comprises 21837 clinical 12-lead ECG records of 10 seconds in length from 18885 patients. For the diagnostic labels, it provides a hierarchical organization in terms of 5 coarse superclasses and 24 sub-classes for the diagnostic labels
Columns provided in the metadata table ptbxl_database.csv. Each ECG is identified by a unique ID (ecg_id) and comes with several ECG statements (scp_codes) that can be used to train a multi-label classifier that can be evaluated based on the proposed fold assignments (strat_fold).

Demographic overview of patients in PTB-XL.

Food Image Classification

Mon, 01 Jan 2024 00:00:00 +0000

This project tackles the challenge of applying computer vision to emphasize accurate food image classification. It compares Convolutional Neural Networks (CNN) with Traditional Machine Learning Models, leveraging CNN’s feature-rich capabilities and traditional models’ transparency and computational efficiency. This dual approach enhances food image classification’s precision and efficiency, impacting quality assessment, dietary analysis, and food inventory management

Used the 5K image dataset for classifying the food images and non-food images, Trained the neural network on the images for classification, and developed the model with streamlit application and achieved 80% accuracy. The dataset is taken from https://mmspg.epfl.ch/food-image-datasets and the name of the data set is 5K.

Dataset description: -. This dataset contains 2500 food and 2500 non-food images, for the task of food/non-food classification in our paper “Food/Non-food Image Classification and Food Categorization using Pre-Trained GoogLeNet Model”. The whole dataset is divided into three parts: training, validation and evaluation. The naming convention is as follows: -. {ClassID}_{ImageID}.jpg -. ClassID: 0 or 1; 0 means non-food and 1 means food. -. ImageID: ID of the image within the class

Bloom's Taxonomy Level Detection

Wed, 13 Dec 2023 00:00:00 +0000

To know Bloom’s Taxonomy in Detail see this pdf This

This project aims to classify the given statement into accurate Bloom’s taxonomy levels out of six levels. This is an NLP-based project. This project is built with TF-IDF vectorizer and multinomial Naive Bayes techniques.

Bloom’s Taxonomy Levels: Bloom’s Taxonomy is a classification system that arranges learning objectives into six hierarchical levels, each representing a different cognitive skill.

The model is trained on 1800 augmented statements. Initially, I had 100 statements for each class, so a total of 600 statements for six classes. Then, this data is augmented three times using contextual word embedding augmentation, and the model used in the augmentation is BERT-based. An example of the augmentation methodology.

This project follows MLOps techniques well. I have divided this project into 6 different pipelines for easy implementation and scalability. I created a class in the pipeline for each class using different entities and components. In generating a configuration file from the user input in the YAML format.

In the end, I created a Docker image so it can be used by the Higher Education Institute(HEIs) for those who are using the OBE(Object Based Education) new curriculum technique. It just reduced the manual process of identifying each question and its root taxonomy level.

Smart Loan Model

Tue, 18 Apr 2023 00:00:00 +0000

This model is trained on a loan dataset with parameters like monthly income, gender, marital status, self-employment status, property area, etc. This model uses all the real-life factors and parameters to detect whether a loan should be given to a person or not. It helps banking or loan departments make an informed decision.

In this project, the models used for classification are logistic regression, K-nearest neighbors classifier, support vector machine, and decision tree classifier.

In this project, the following steps were taken: data collection, data preprocessing, data splitting, model implementation, and model evaluation. A detailed description of each step is given below:

Data Collection: I collected the data from Kaggle and read it with the pandas library. I analyzed the data frame using df.info() and df.describe().
Data Preprocessing: For preprocessing, I dropped the duplicate values and removed and filled the null values with their respective methods, like using the median for numerical data and the mode for objective data. For visualization, I drew some plots like line plots, countplots, and bar plots for counting the loan status employment status, and other attributes to get a brief understanding of the dataset. I also generated the categorical and numerical data separately for further splitting.
Data Splitting: In this state, I divided the data into 80% for training and 20% for testing.
Model Implementation: I created a dictionary where the keys are the model names, and the values are the model instances themselves with parameters. I applied functions like loss calculation and model evaluation to this dictionary. Output: Finally, I saved the data in a suitable format using pandas.