I’m getting quite obsessed with the AI Revolution stuff and thought it could be worth starting a thread to share resources which could be applied to ME in some way. I know some of the ME researchers do keep up with the forum. It might be that some others on the forum have skills to use existing research data and analyse it differently with the advent of open source AI tools? The first is https://owkin.com/substra/
https://consensus.app/search/ Ask questions of research papers and also get a synthesise summary But PACE stretches its evaluation limits:
Possibly of relevance to research around POTS — Artificial intelligence-based model to classify cardiac functions from chest radiographs: a multi-institutional, retrospective model development and validation study (2023, The Lancet Digital Health)
Sorry I didn’t keep this thread going. Are there similar threads? I’m assuming you’re probably using ChatGPT scheduled task feature or similar to keep up with relevant research these days? I can’t read very much so I find the Google lm notebook podcast feature useful when there’s new ME papers. Does anyone know if anyone has access to this and ME blood samples: Mal-ID (Machine Learning for Immunological Diagnosis) combines six machine learning models to analyze millions of immune cell sequences, identifying distinct patterns associated with various diseases[1][2]. This groundbreaking approach examines both B-cell and T-cell receptors (BCRs and TCRs), with B cell receptor sequences proving most effective for detecting HIV and SARS-CoV-2 infections, while T cell receptor sequences provide better insights into lupus and Type 1 diabetes[3]. The tool's combined analysis improves diagnostic accuracy across all conditions, regardless of patient demographics, and can even detect recent flu vaccinations[1][4]. Sources I got information from https://www.perplexity.ai/page/ai-tool-diagnoses-diabetes-hiv-60yD.7CfT9OBJzcTZ.LYMw which cites: [1] Disease diagnostics using machine learning of B cell and T cell receptor sequences https://www.science.org/doi/10.1126/science.adp2407 [2] Disease diagnostics using machine learning of B cell and ... - Science https://www.science.org/doi/abs/10.1126/science.adp2407 [3] Immune ‘fingerprints’ aid diagnosis of complex diseases in Stanford Medicine study https://med.stanford.edu/news/all-news/2025/02/immune-cell-receptors-complex-disease.html [4] AI tool diagnoses diabetes, HIV and COVID from a blood sample https://www.linkedin.com/posts/dr-e...v-and-covid-activity-7298783474461605889-ubLc The innovative "one-shot sequencing method" employed by Mal-ID captures comprehensive immune system exposures, providing a holistic view of an individual's health status. This approach allows for the simultaneous assessment of multiple diseases through a single blood test, streamlining the diagnostic process[1][2]. By analyzing millions of immune cell sequences, the system can detect subtle patterns indicative of various conditions, offering a more nuanced understanding of a patient's immune response[3]. This method's ability to provide a unified immune system analysis represents a significant advancement in diagnostic medicine, potentially reducing the time and resources required for accurate disease identification[4]. Sources [1] AI tool diagnoses diabetes, HIV and COVID from a blood sample https://neuron.expert/news/ai-tool-diagnoses-diabetes-hiv-and-covid-from-a-blood-sample/11113/en/ [2] Disease diagnostics using machine learning of B cell and T cell receptor sequences https://www.science.org/doi/10.1126/science.adp2407 [3] Disease diagnostics using machine learning of B cell and ... - Science https://www.science.org/doi/abs/10.1126/science.adp2407 [4] Machine Learning Unlocks Immune System Secrets https://www.insideprecisionmedicine...chine-learning-unlocks-immune-system-secrets/
Thanks for suggesting using Notebook LM podcast feature to access research studies. I've found that particular feature very useful for other things, but hadn't thought of using it on research papers. I've found Gemini very useful when writing letters etc, I can put in a rough paragraph and it comes up with something coherent for me. I was looking at a long list of free text patient feedback yesterday. I asked it to identify positive and negative themes, and it does it in a second or two. Obviously responses need to be sanity checked, but in the last six months or so the AI available without subscription has improved enormously.
