Data Management & Statistics Jobs in Life Sciences’ Post

Back on November, I assisted an inspiring conference hosted by the Flatiron Health Data Science team on machine learning and cancer, it's because of this that I wanted to challenge myself and create a Breast Cancer identification and diagnostic model. Here's the full Readme and the link to the repository! https://lnkd.in/djFeHUch 🔬 Project Goal: To develop a model that can diagnose breast cancer with accuracy, using a dataset of breast cancer patient records. 📊 Dataset and Approach: The project utilized digitized images of fine needle aspirate (FNA) of breast masses, focusing on cell nuclei characteristics to differentiate between benign and malignant diagnoses. Through a meticulous process involving data preprocessing, exploratory data analysis, feature engineering, and model optimization, I was able to create a model that could potentially save lives by enabling early detection. 💻 Technological Framework: Employing Python, Jupyter Notebook, and libraries like Pandas, NumPy, Matplotlib, Seaborn, and Scikit-Learn, the project was a deep dive into the capabilities of machine learning in medical diagnostics. 🔍 Conclusions and Results: The journey culminated in the development of an optimized Random Forest model that boasts a 96% accuracy rate in diagnosing breast cancer. This achievement underscores the potential of integrating machine learning with healthcare to enhance diagnostic processes significantly. 🌐 Looking Ahead: The next steps for this project include expanding the dataset for greater robustness, exploring more advanced algorithms, and developing a web application to make this tool accessible to healthcare professionals globally. Please check the repository and feel free to make any comments or insights about the project, I would also love to collaborate with other programmers in the future to create more projects :) #DataScience #MachineLearning #BreastCancerAwareness #HealthcareInnovation #FlatironHealth

Small things that make me very happy. In a recent study characterizing macrophage heterogeneity in lymphomas, the authors used our HCATonsilData package to access the myeloid cells from our tonsil atlas and look at the expression of their favourite genes. Sharing the data of your papers is an exercise of EMPATHY with the scientific community. It's not so different from the empathy you need with your audience when preparing a talk. Who is your target audience? What is their background? What do they need to understand the talk? Similarly: - Future studies might want to integrate your dataset with others and process all of them in a uniform pipeline. Therefore, you need to provide your raw reads and document well how to map them, as well as curating well your metadata. Sometimes, however, this data might be sensitive and thus need to be protected under restricted access. - Other users might want to avoid the hassle of remapping all the data and want to start from the count matrices directly, perhaps revisiting the QC thresholds you applied - Other users might want to have a quick look to your data to generate hypothesis or validate a finding they have derived from another dataset. Providing your processed R or Python objects can be a game-changer. In our manuscript, we went a step further and developed an R package that provides access to the most comprehensive atlas of cells from a secondary lymphoid organ to date. Users can easily query they favourite data modality in their favourite cell lineage with as few as 3 lines of code. We take pride in the FAIRness of our data: findable, accessible, interoperable, and reusable. Our atlas is not just "stamp collection" or "a UMAP with labels". It's a resource that allows you to generate hypothesis, trace back annotations, and interpret signatures from lymphoid tumors. We still have room for improvement. Last Friday I presented our work to an experimental lab focused on B cell lymphomas. Multiple members of the team don't have coding experience. We willnow also work to make it fully accesible through the browser.

🔍 Breast Cancer Detection Project: Exploratory Data Analysis and Model Building 🔍 Exciting news! 🌟 Fresh off completing the Exploratory Data Analysis for Machine Learning course on IBM, I dove into a meaningful project focusing on Breast Cancer Detection. Breast cancer affects millions worldwide, making early detection pivotal. Here's a glimpse into the key steps taken in this project: 1️⃣ Data Loading and Cleaning: We start by loading the breast cancer dataset, performing data cleaning, and understanding its structure using descriptive statistics. 2️⃣ Binary Encoding: Categorical variables like "M" (Malignant) and "B" (Benign) are converted to numerical values (1 and 0) for compatibility with machine learning algorithms. 3️⃣ Feature Selection: Using correlation analysis and heatmap visualization, we identify the most relevant features that explain the relationship between independent variables and the target variable (diagnosis). 4️⃣ Visualization: We create pair plots and correlation matrices for better visualization of relationships between selected features, helping us understand patterns and potential insights. 5️⃣ Feature Scaling: Normalization and standardization are applied to ensure all features have the same scale, a crucial step for many machine learning algorithms. 6️⃣ Handling Outliers: Outliers are identified and removed to enhance the robustness of our model. 7️⃣ Feature Extraction with PCA: Principal Component Analysis (PCA) is employed to reduce dimensionality while retaining relevant information. We then build a logistic regression model using the extracted features. 8️⃣ Model Evaluation: Our model achieves an accuracy of 97.37% on the test set, demonstrating its effectiveness in classifying breast tumors. 🚀 Conclusion: The EDA process lays the foundation for a robust breast cancer detection model. Continuous improvement and refinement can be made based on additional insights and model evaluation metrics. I'm thrilled to share that the project has been uploaded on GitHub! Feel free to explore the code and share your thoughts!, and let's contribute to advancing breast cancer detection through data-driven approaches. 🌸💻 👉 GitHub Link: https://lnkd.in/dnJVjX3r #DataScience #BreastCancerAwareness #MachineLearning #DataAnalysis #IBM #machinelearningalgorithms #featureengineering

Here are my 12 guidelines for data exploration and analysis with the right attitude for discovery - Itai Yanai #Cancer #DataExploration #Datasets #OncoDaily #Oncology 

🚀 Exciting News in Data Science! 📊 I'm thrilled to share the results of my latest project, an in-depth Exploratory Data Analysis (EDA) comparing the efficacy of Capomulin, an anti-cancer medication, against other treatment regimens. 🌐✨ Project Highlights: Conducted a comprehensive inter-drug comparison to evaluate treatment effectiveness. Explored key metrics and visualizations to gain insights into drug performance. Dived deeper into Capomulin's impact, including a detailed linear regression analysis. Examined the relationship between sample weight and tumor volume to understand potential correlations. Explore the Project: https://lnkd.in/gHqpfUJz I'm proud of the progress made and the valuable insights gained throughout this journey. Let's continue to advance data-driven discoveries together! 💡📈 #DataScience #Python #Matplotlib #LinearRegression #ExploratoryDataAnalysis #CapomulinStudy #DataAnalytics #ProjectShowcase #LinkedInProject

Knowing full well that this exposé is a work in progress I could highlight two good observations it makes of (1) that outlier 'variants' potentially harbour alternate truths and (2) that the dominant may not be well positioned. We hope to get a reconciliation procedure in the paradigm being suggested here. One such would be to not rush to press on the first finding but to return to each of the variants to test the finding again as though it was not a 'variant'. That would mollify my usual Type II quibbles I always throw at strident AI.

"Excited to share my latest visualization on breast cancer data analysis using Tableau! 📊 Exploring the intricacies of this vital dataset has been both enlightening and inspiring. From uncovering patterns to identifying potential predictive factors, every insight gained brings us closer to better understanding this complex disease. Grateful for the opportunity to leverage data analytics for meaningful impact in healthcare. #DataAnalytics #Tableau #BreastCancerAwareness #HealthcareAnalysis"

How far can you go with Alteryx SparkED? Spenser Johnson brought his passion for scientific research and his drive to work with data to a cancer research lab at Yale University. Here Spenser uses Alteryx solutions to pursue ground-breaking cancer insights. As a graduate researcher, Spenser is able to use a free Alteryx license through the SparkED program. Read the amazing story of how Spenser is using #dataanalytics to develop more precise cancer treatments, and learn why #analytics and #medicalresearch are a perfect match. http://ow.ly/1yp310528AO #alteryx #alteryxsparked #analyticsforall

Exploring Breast Cancer Detection: My Data Journey I'm thrilled to share a glimpse into my latest project where I delved into the world of breast cancer detection using data analysis techniques. Armed with a dataset featuring parameters like diameter, radius mean, texture mean, and more, I embarked on a journey to uncover insights and predictions that could potentially make a difference in healthcare. From harnessing the power of random forest and logistic regression to crafting informative confusion matrices, every step of this journey was a learning experience. It's fascinating how data science can empower us to make informed decisions and even contribute to life-saving discoveries. While the technical aspects were certainly challenging, what truly resonated with me was the potential impact this work could have on real lives. Being able to leverage data in such a meaningful way reaffirms my passion for this field and reinforces my commitment to driving positive change through analytics. I'm incredibly proud of what I've accomplished with this project, and I'm excited to continue exploring the intersection of data science and healthcare. Here's to using data for good and making a difference one analysis at a time! #DataScience #BreastCancerDetection #Analytics #Healthcare #DataForGood #LinkedInPost #ml #cancerawareness #cancercare

2024 is The Year of Data Scientist Alhamdulillah, starting in 2024, I have completed my studies in Data Science after previously taking several short courses and learning independently on YouTube, forums, etc. I am immensely grateful to Coach Mega Bagus Herlambang for all his knowledge and experience sharing. My initial independent project, developed from scratch, involves creating a Machine Learning-based application for "Breast Cancer Prediction", to determine whether it is still in a malignant (dangerous) stage or benign (safe) phase. 1. I obtained the dataset from https://lnkd.in/g7HDP-_N. The process of creating the model flow was done independently, without following any source code from dataset author. 2. From the original data, there are 31 features used for creating the machine learning model. However, for the deployment process, I added an initial "preprocessing" step where I extracted only the 10 most important features to simplify the input data process. 3. Through the evaluation of several models, the best performance was obtained using the Ada Boost Classifier model. 4. In the processing phase, it was identified that there was an indication of an imbalanced dataset, leading to the implementation of an over-sampling process to approach a balanced dataset. Here is the link of deployement result created using the Streamlit platform: https://lnkd.in/gZswnTJ7