Harini Gopalakrishnan

Interesting that on the MIMIC IV readmission models, the LLMs get closer to the traditional ML models. Perhaps because LLMs are not great at representing numeric values (especially at the model sizes tested) that are informative of the prediction for acute outcomes, whereas readmission relies more on diagnoses and other more linguistically-coded inputs. (Though that doesn't explain the poor performance on MIMIC-III readmission)

3

This California Management Review article explores the ROI of AI ethics and governance investments through two lenses: loss aversion and value generation. While many companies focus on minimizing risks like legal penalties and reputational harm, the article argues that a value generation approach—using ethics to build trust, foster innovation, and create competitive advantages—offers greater long-term benefits. By integrating ethics into strategic planning, organizations can both mitigate risks and enhance overall growth.

49

3 Comments

Organizations have insufficient technical capabilities for the speed and scale of their ML model implementations. My colleague Sumit Agarwal just published a very helpful research with a capability view and sample vendors for defining a modular, flexible and scalable MLOps solution. https://lnkd.in/gnB_T9bq #MLOps #LLMOps #AI

149

17 Comments

As #ForrTech kicks off, Forrester Analyst Jayesh Chaurasia is interviewed by InformationWeek about data governance and how #AI strategies rely on a foundation of strong data capabilities to capture and deliver data at scale for AI solutions to support an organization. #technology

6

From Demos to Deployment: Navigating Challenges of Scaling AI Solutions Good overview on AI in healthcare, where early adopters share their frustrations with current AI solutions' limited ROI and effectiveness in clinical practice. Despite 96% of respondents being underwhelmed, the future looks brighter with advancements like large language models and FDA-cleared AI medical devices. The integration of AI into healthcare holds immense promise if done thoughtfully! #airevolution #artificialintelligence #largelanguagemodels

18

3 Comments

👉🏼 How effective are large language models in product risk assessment? 🤓 Zachary A Collier 👇🏻 https://lnkd.in/eZy9nVrh 🔍 Focus on data insights: - Large language models, like ChatGPT, show promise in brainstorming potential failure modes and risk mitigations. - However, there were errors and inconsistencies in some results. - The guidance provided by the models was perceived as overly generic and occasionally outlandish. 💡 Main outcomes and implications: - ChatGPT performed better at divergent thinking tasks but lacked depth of knowledge compared to subject matter experts. - Similar patterns in strengths and weaknesses were observed when comparing ChatGPT to other large language models. - Despite challenges, there may be a role for large language models in assisting with ideation in product risk assessment. 📚 Field significance: - Product safety professionals can leverage large language models for brainstorming and idea generation in risk assessment processes. - Experts should focus on critically reviewing AI-generated content to ensure accuracy and relevance. 🗄️: [#productrisks #AI #riskassessment #ChatGPT]

1

1 Comment

We're proud to share that Northwestern Medicine is implementing Nuance's DAX Copilot, embedded in Epic, across its entire enterprise! This cutting-edge ambient voice solution, powered by Microsoft Cloud for Healthcare, is designed to ease the administrative load on physicians, enabling them to dedicate more time to what matters most—patient care. With DAX Copilot, Northwestern Medicine is enhancing both physician satisfaction and patient experiences, demonstrating their commitment to leveraging advanced AI for better health outcomes. Discover how DAX Copilot is driving innovation in healthcare: https://lnkd.in/dRg7NKhE  #Healthcare #MSFTAdvocate #DAXCopilot

17

GenAI for #HEOR and #RealWorldEvidence , a must read for #aiforlifescience

24

Do we need data standards when we have LLM? Data we need to train our AI models at Pacmed are far from standardized, they differ on many dimensions: data differ greatly between EHR systems since they have completely different data structures, they differ between versions of EHR systems, they differ between hospitals using the exact same EHR system and EHR version and they differ within a hospital over time. An important step we take in implementing our Pacmed Critical solution is therefore processing, enriching and harmonizing data of the ICU. This also involves working closely with data and clinical experts within the hospital to validate the data. This is the only way we can ensure the data we use for the AI models are valid and produce high quality and meaningful predictions. With the introduction of large language models one could argue this will make the task much easier or even would make it unnecessary to involve IT and medical experts. Our medical data specialist Arjen Mol discusses his point of view based on a recent article in NEJM AI. What is your perspective?

9

When the marketing team knows how to have fun 😂 🔥 . What it looks like in practice: ➡ Identifying *specific* groups who struggle with common depression treatments ➡ Distinguishing between *types* of depression for better care (we don't say all cancer is the same, why do chronic disease patients get lumped under one umbrella?) ➡ Clarifying the *spectrum* of depression for targeted therapies When you combine deep data with physician intuition and (6x!) patented AI tools, you can do some sci-fi level science. #JustSayin' #FunWithData #MemesAndMachineLearning

4

👉🏼 Battle of the bots: a comparative analysis of ChatGPT and Bing AI for kidney stone-related questions 🤓 Amber K McMahon 👇🏻 https://lnkd.in/efETTsfa 🔍 Focus on data insights: - 📊 ChatGPT demonstrated superior performance in providing clear, relevant, and appropriate responses based on AUA guidelines compared to Bing AI. - 🔍 While ChatGPT scored higher in clarity and relevance, Bing AI offered references, which aids in assessing information quality. - ⚖️ Significant variations in adherence to clinical guidelines were noted, with both chatbots showing instances of recommendations against established guidelines for certain scenarios. 💡 Main outcomes and implications: - 🏆 ChatGPT's ability to generate clearer and more relevant responses suggests it could be more effective for patient education and preliminary inquiries in urology. - 📚 The inclusion of references by Bing AI highlights the importance of source credibility in medical queries, suggesting a need for improvement in reference utilization for conversational AI models. - 🧠 The findings underscore the necessity for ongoing evaluation of AI tools in healthcare to ensure they align with clinical guidelines and provide reliable information. 📚 Field significance: - 🤖 The study contributes to the growing body of literature on AI applications in healthcare, emphasizing the strengths and weaknesses of existing models. - 🩺 It encourages future research into AI's role in clinical settings, particularly regarding its integration in patient interactions and education. - 🌐 The implications of these findings advocate for enhanced designs in chatbot interfaces to optimize user experience while maintaining adherence to clinical standards. 🗄️: [#AI #ChatGPT #BingAI #Urology #KidneyStones #PatientEducation #ClinicalGuidelines #HealthcareTechnology #InformationQuality]

1

Don't miss to enroll into this session to learn about how data governance is embedded into Design and Architecture of watsonx.data. #watsonx #data #information #watsonx.data #IBM #AI

75

2 Comments

Multi token prediction from AI at Meta - a significant advancement in #LLM. Rather than the traditional approach of predicting the next word in the sequence, the recently published paper and the corresponding model on Hugging Face forecasts multiple future words simultaneously, promising enhanced performance and drastically reduced training times. The novel approach from AI at Meta could have profound positive implications - lowers the cost and environmental impact of LLMs - Significant improvement in tasks from code generation to creative writing And also the downside of increasing the potential for misuse particularly with the AI at Meta strategy of open licensing through Hugging Face Check out the research paper here: https://lnkd.in/eAP2RufM

52

2 Comments

Exploring the intricacies of implementing Retrieval Augmented Generation (RAG) use-cases in healthcare? Read my latest article that delves into the unique challenges and advanced solutions that are shaping the future of AI in healthcare front-office applications. #healthcare #ai #genai #rag

46

3 Comments