© 2024 Michael Truppe, MD
MedlibreGPT: Transforming Healthcare by Integrating Secure Patient Records and Specialized Medical Literature into PrivateGPT Models on Open-Source Nextcloud Servers.
Our Mission
MedlibreGPT is revolutionizing healthcare by integrating advanced AI in teleconsultation. We combine PrivateGPT models with medical literature expertise to meet diverse medical needs. Our focus includes secure patient record integration, maintaining confidentiality and trust.
We aim to provide healthcare professionals with advanced insights and data-driven tools, using AI on Nextcloud servers. MedlibreGPT represents a significant leap in healthcare quality, efficiency, and accessibility.
Our continuous innovation and dedication aim to establish new standards in healthcare technology. At the heart of our mission is the ethical use of technology for superior patient care, positioning MedlibreGPT at the forefront of medical teleconsultation.
Project on Github
Research Projects
Artificial Intelligence for Detection of Oral Cancer
Using past medical records and current literature to create predictive models.
AIDOCDies wäre unser aktuelles Projekt
AI-CARES: Enhancing Teleconsultation through AI-Driven PersonalizationBURNCAREBURNCARE webBURNCARE padEnhancing Post-Discharge Dental Implant Care through AI-Powered TelemonitoringMedlibreGPT Healthcare Data Management
Healthcare Data Management and Medical Research
Objective
The MedlibreGPT project is dedicated to revolutionizing healthcare through the development of proprietary PrivateGPT models. These advanced AI models are meticulously designed to synergize with specialized medical literature, catering to the nuanced needs of various highly specific medical fields.
Core Concept
At the heart of MedlibreGPT is a robust integration of secure patient records with a rich knowledge database of medical literature. This integration empowers healthcare professionals with tailored insights and data-driven decision-making capabilities.
Technology Stack
We leverage the power of open-source Nextcloud servers, ensuring that all patient records are securely stored and managed on-premises. This approach not only enhances data security but also maintains the integrity and confidentiality of patient information.
Vision
Our vision is to set a new standard in healthcare information management. By combining advanced PrivateGPT models with accessible and secure data storage solutions, MedlibreGPT aims to transform the landscape of medical research and patient care.
Questions? Here’s how to reach us:
- Slack channel: projekt-ai-graz
- Slack Email: projekt-ai-graz-aaaalvw6oofdcozs32uqik7ffm@medlibre.slack.com
- Team manager: Michael Truppe, MD
Open Source Benefits for MedlibreGPT
- Accessibility and Collaboration: MedlibreGPT's use of open-source Nextcloud servers facilitates broader access to advanced healthcare technology. This approach promotes global collaboration, inviting healthcare professionals and developers to contribute, enhancing the technology's evolution and keeping it at the forefront of teleconsultation.
- Customization and Flexibility: The open-source nature of MedlibreGPT allows healthcare providers to tailor the platform to their unique needs. This adaptability is vital in addressing diverse medical challenges, making the technology not only advanced but also relevant across various healthcare scenarios.
- Transparency and Trust: Committing to open-source, MedlibreGPT prioritizes transparency. This allows users to understand and trust the AI models' workings, which is essential in maintaining ethical standards in sensitive areas like patient records and medical decision-making.
- Cost-Effectiveness: Open-source solutions like MedlibreGPT are often more affordable, lowering barriers for resource-limited healthcare providers. This aligns with the goal of enhancing healthcare quality and accessibility globally, particularly in underserved areas.
AContinuous Improvement: The open-source model encourages ongoing feedback and updates from a worldwide user and developer community. This constant evolution is crucial in healthcare, a field where keeping up with the latest research and patient care strategies is key.
Unique advantages of MedlibreGPT
AI Physician Twin
We create sophisticated virtual counterparts of physicians, equipping these AI twins with an extensive knowledge base derived from a comprehensive corpus of medical literature. This integration endows the virtual entities with advanced capabilities, reflecting and potentially surpassing the skill set of their human counterparts.
Virtual Patients on Nextcloud
We have created virtual patients, complete with comprehensive medical histories, to evaluate the system's responsiveness and potential biases. Regular patients are securely onboarded using two-factor authentication (2FA) and are provided with a personal identification number (PIN) via SMS, which grants them access to their personal data. MedlibreGPT accesses individual patient data on Nextcloud.
Front End to MedlibreGPT
Get a treatment proposal via SMS and Email in a few minutes. We have different front end versions and different physician skills as AI twins. You can access also via IBM Watson (https://watson.smile.wien).
You are free to test the system. You will be asked for email, name and mobile number at the end. The SMS will work internationally.
https://link.truppe.at/meinlachen
Sample MedlibreGPT answer to patient via SMS and EMAIL
The documents are organized into various sections, each crafted by distinct AI entities representing different medical specialties. Our objective is to form an integrated virtual treatment team, encompassing professionals from a variety of medical disciplines.
Additionally, we are conducting an evaluation of how brand names are contextually mentioned within the content generated by these AI entities
Sample Questions to MedlibreGPT
We are undertaking an assessment to examine how the PrivateGPT RAG system handles references to literature. This evaluation specifically focuses on the system's response to queries pertaining to distinct sections within the processed documents.
The current testing approach utilizes OpenAI's technology, primarily due to its rapid processing capabilities. However, we are transitioning to a local Large Language Model (LLM) solution. This shift is particularly crucial as it involves the integration of individual patients' medical histories, which are securely stored on a Nextcloud server. This change underscores our commitment to enhancing both data security and processing efficiency.
Australian Outback Adventure Turns Urgent: Burn Wound and Hyperglycemia Concerns for Diabetic TravelerDermatologist on Australian Outback Adventure Seeks Advice for Treating a 3cm Diameter Deep Burn WoundResearch Grants
Project P 12464 The Optical Interface for Augmented Reality in Computer Assisted Navigation and 3D Visualization in Surgery
Dr. Truppe’s research has been supported by the FWF (Austrian Science Fund, Project P 12464 The Optical Interface for Augmented Reality in Computer-assisted Navigation and 3D Visualization in Surgery; http://dx.doi.org/10.13140/RG.2.1.3117.8403) and DFG (German Research Foundation). His patents and scientific publications have been cited over 2700 times worldwide (https://link.truppe.at/googlescholar).
DFG Project: D. Siebert, “Ein Meßsystem zur präoperativen Planung und intraoperativen Kontrolle von Dysgnathieoperationen,” Medizinische Fakultät Charité – Universitätsmedizin Berlin, Medizinische Fakultät Charité, Berlin, 2002.
FWF Project: PDF download P 12464
DGF Project Dissertation D. Siebert
Wayback Machine
The Wayback Machine is a digital archive of the World Wide Web.
Discord
This is an open source initiative. Contributions are welcome.