Posts

Establishing Causality in product management

# The 9 Bradford-Hill Criteria for Establishing Causality: Industry Examples The Bradford-Hill criteria, established by Sir Austin Bradford Hill in 1965, provide a framework for determining whether an observed association is likely to be causal. These nine criteria have become fundamental tools in epidemiology, but their application extends far beyond healthcare into various industries. This article explores each criterion with three practical examples from different sectors. ## 1. STRENGTH: How large is the association? The strength criterion examines the magnitude of the observed association between a potential cause and effect. **Healthcare Industry:** The exceptionally strong association between smoking and lung cancer (smokers are 15-30 times more likely to develop lung cancer than non-smokers) was one of the first compelling pieces of evidence that smoking causes cancer. **Technology Industry:** A strong correlation exists between screen time before bed and sleep disruption. Stud...
Post a Comment
Read more

AI Agent queries people ask from Chat GPT

 Trends emerging in the types of queries people have about implementing AI. Here are a few common patterns: 1. Industry-Specific Use Cases Healthcare : How to deploy AI for diagnostics, personalized medicine, or patient record management. Manufacturing : Optimizing production lines with predictive maintenance and quality control. Retail : Personalization engines, dynamic pricing, and supply chain optimization. Finance : Fraud detection, portfolio optimization, and customer service bots. Education : AI tutors, automated grading, and adaptive learning systems. 2. Infrastructure and Tools Many ask about tech stacks for AI implementation, such as using PyTorch or TensorFlow for training models, and AWS, Azure, or Google Cloud for deployment. Questions about scalability , like ensuring the AI system handles large loads effectively. 3. Ethics and Regulations Queries focus on ensuring AI systems are fair, transparent, and explainable . Compliance with privacy laws (e.g., GDPR, HIPAA). 4....
Post a Comment
Read more

IoT Technologies , Protocols, stacks, resources

A. Examples of IoT applications from different industries for each specified element: Actuators Industry: Automotive Example: Automatic Braking Systems in Vehicles Actuators are used in automatic braking systems to apply pressure to the brake pads when a collision risk is detected. Sensors monitor proximity, speed, and environmental changes, and the actuator responds by engaging the braking mechanism to prevent a collision. Embedded Systems Industry: Healthcare Example: Wearable Health Monitoring Devices Embedded systems are used in wearable devices like smartwatches or fitness trackers. These devices monitor vital signs (e.g., heart rate, blood pressure) using sensors and process the data using microcontroller-based embedded systems to provide insights into a user’s health in real-time. Intelligent Devices Industry: Agriculture Example: Smart Irrigation Systems Intelligent devices in smart agriculture use IoT capabilities to optimize irrigation schedules. These devices compute soil mo...
Post a Comment
Read more

Medical device Common Technical Document (CTD) application. US and EU

 The Common Technical Document (CTD) is a standard format for presenting data in an Investigational New Drug (IND) application to the FDA Includes a good Graphic. How to handle your Medical Device Technical Documentation? – Part 1: Basic definitions https://www.debiotech.com/how-to-handle-your-medical-device-technical-documentation-part-1-basic-definitions/ From FDA :  (a) Electronic Common Technical Document (eCTD) https://www.fda.gov/drugs/electronic-regulatory-submission-and-review/electronic-common-technical-document-ectd (b) Guidance document. FAQs. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/m4-ctd-general-questions-and-answers Ref https://www.bioduro.com/news-resources/insights/common-technical-document.html#:~:text=The%20Common%20Technical%20Document%20%28CTD%29%20is%20a%20standard,ensuring%20consistent%20and%20comprehensive%20information%20for%20regulatory%20review. Additionally: How to CE Mark a Medical Device that Incorporates a Dru...
Post a Comment
Read more

Career routes and examples of professionals progressing to AI/ML professional

Reaching the level of an AI/ML Architect specializing in Generative AI typically requires a combination of strong technical expertise, continuous learning, and career progression through increasingly complex roles in AI and data science. Below are some common career routes and realistic examples of professionals progressing to this level : Career Routes 1. Academic to Industry Path Phase 1: Education and Research Earn a Bachelor's/Master's in Computer Science, Mathematics, or a related field. Pursue a Ph.D. in AI, ML, or NLP with a focus on generative AI topics such as transformers, GANs, or large language models. Publish research papers or contribute to open-source AI projects. Phase 2: Entry into Industry Begin as a Research Scientist or AI Engineer in a research-driven organization or lab. Phase 3: Senior Roles Transition to senior roles like Lead ML Engineer or Principal Data Scie...
Post a Comment
Read more

JD: AI ML senior software professional / architect

JD of: AI/ML Architect with expertise in Generative AI to develop and deploy advanced machine learning models that solve real-world business challenges. The ideal candidate should have a strong understanding of transformer architectures, a hands-on approach to handling large datasets, and the ability to scale solutions in production environments. -----  Responsibilities Model Development: Build and fine-tune generative AI models such as GPT, BERT, T5, GANs, VAEs, and autoregressive models. Experiment with multimodal AI models combining text, images, and other data modalities. Data Handling and Preprocessing: Collect, clean, preprocess, and transform large and complex datasets for model training. Perform tokenization, text normalization, and feature extraction for unstructured data. Automate data preprocessing pipelines using Python scripts or ETL frameworks. Model Optimization and Deployment: Implement techniques like transfer learning, fine-tuning, and hyperparameter optimization ...
Post a Comment
Read more

Case : OPEN AI to do analysis and calculation for a report

 Used this prompt to get my work done: Input ( copied data from a PPT and pasted after the prompt): based on frequency percentage , divide number 5 into each item.:   Incorrect DOS split –phase 1-PH​ In progress​ 5%​ Incorrect DOS split –phase 2-PH​ In backlog​ 6%​ Incorrect code prediction -AI MODEL stage-PH​ In backlog​ 4%​ Inaccuracies in valid & invalid, incorrect document identification-PH​ In backlog​ 5%​ Incomplete/incorrect logic applied for MEAT/TAMPER-PH​ In backlog​ 4%​ Issue AI output/AI MODEL prediction-PH​ In backlog​ 3%​ Relevant DOS logic-PH​ In backlog​ 5%​ Section split issue-PH​ In backlog​ 3%​ Output: To distribute the number 5 based on the given percentages, we can calculate the share for each category by multiplying 5 by each percentage and then rounding to an appropriate precision. Here’s the calculation: Categories and their percentages Incorrect DOS split – Phase 1-PH : 5% Incorrect DOS split – Phase 2-PH : 6% Incorrect code prediction - AI MO...
Post a Comment
Read more