Medical Chatgpt Save

Explorations into training a ChatGPT, but tailored towards primary care medicine, with the reward being able to collect patient histories in a thorough and efficient manner and come up with a differential diagnosis. May also explore to see if it can be further fine-tuned on pirated copies of Up-To-Date for specialist knowledge

Sadly, I no longer think this is possible in its current state. It will probably see some utility with scribing of basic bread and butter cases; however assess and plan it cannot.

Citations

@inproceedings{Singhal2022LargeLM,
    title   = {Large Language Models Encode Clinical Knowledge},
    author  = {Karan Singhal and Shekoofeh Azizi and Tao Tu and Said Mahdavi and Jason Lee Kai Wei and Hyung Won Chung and Nathan Scales and Ajay Kumar Tanwani and Heather J. Cole-Lewis and Stephen J. Pfohl and P A Payne and Martin G. Seneviratne and Paul Gamble and Chris Kelly and Nathaneal Scharli and Aakanksha Chowdhery and P. D. Mansfield and Blaise Ag{\"u}era y Arcas and Dale R. Webster and Greg S. Corrado and Y. Matias and Katherine Hui-Ling Chou and Juraj Gottweis and Nenad Toma{\vs}ev and Yun Liu and Alvin Rajkomar and Jo{\"e}lle K. Barral and Christopher Semturs and Alan Karthikesalingam and Vivek Natarajan},
    year    = {2022}
}

@article {Kung2022.12.19.22283643,
    author  = {Kung, Tiffany H. and Cheatham, Morgan and , and Medenilla, Arielle and Sillos, Czarina and De Leon, Lorie and Elepa{\~n}o, Camille and Madriaga, Maria and Aggabao, Rimel and Diaz-Candido, Giezel and Maningo, James and Tseng, Victor},
    title   = {Performance of ChatGPT on USMLE: Potential for AI-Assisted Medical Education Using Large Language Models},
    elocation-id = {2022.12.19.22283643},
    year    = {2022},
    doi     = {10.1101/2022.12.19.22283643},
    publisher = {Cold Spring Harbor Laboratory Press},
    URL     = {https://www.medrxiv.org/content/early/2022/12/21/2022.12.19.22283643},
    eprint  = {https://www.medrxiv.org/content/early/2022/12/21/2022.12.19.22283643.full.pdf},
    journal = {medRxiv}
}

@misc{https://doi.org/10.48550/arxiv.2301.10035,
    doi     = {10.48550/ARXIV.2301.10035},
    url     = {https://arxiv.org/abs/2301.10035},
    author  = {Nov, Oded and Singh, Nina and Mann, Devin},
    keywords = {Human-Computer Interaction (cs.HC), FOS: Computer and information sciences, FOS: Computer and information sciences},
    title   = {Putting ChatGPT's Medical Advice to the (Turing) Test},
    publisher = {arXiv},
    year    = {2023},  
    copyright = {Creative Commons Attribution Share Alike 4.0 International}
}

@inproceedings{Schick2023ToolformerLM,
    title   = {Toolformer: Language Models Can Teach Themselves to Use Tools},
    author  = {Timo Schick and Jane Dwivedi-Yu and Roberto Dessi and Roberta Raileanu and Maria Lomeli and Luke Zettlemoyer and Nicola Cancedda and Thomas Scialom},
    year    = {2023}
}

@inproceedings{Peng2023CheckYF,
    title     = {Check Your Facts and Try Again: Improving Large Language Models with External Knowledge and Automated Feedback},
    author    = {Baolin Peng and Michel Galley and Pengcheng He and Hao Cheng and Yujia Xie and Yu Hu and Qiuyuan Huang and Lars Lid{\'e}n and Zhou Yu and Weizhu Chen and Jianfeng Gao},
    year      = {2023}
}

@inproceedings{Nori2023CapabilitiesOG,
    title   = {Capabilities of GPT-4 on Medical Challenge Problems},
    author  = {Harsha Nori and Nicholas King and Scott Mayer McKinney and Dean Carignan and Eric Horvitz},
    year    = {2023}
}