What is Govern Corp, what’s God’s eye?
It’s important first to understand the vision of Govern Corp. It’s to create a God’s Eye, a software terminal that allows everyone within an enterprise operations setting to derive intelligence from large datasets, documents, patient reports, medical data sources allowing for the most human interface with complex system. At the same time, the system should allow staff, doctors, and nurses to launch actions such as sending immediate referrals, changing schedules, monitoring status of beds, information on supply chain, etc. all through a multilingual chat system, that runs on web, iPad, and smartphones.
As humans, our mode of communication is natural language. We engage in all our activities using the English language, however to interact with complex data and systems we require a medium to make that communication easy such as a dashboard that is designed to interact with a set of data or an enterprise software that performs certain functions that you can launch by navigating and clicking the interface. However, all this can be replaced with a chat system that engages not only natural language, but a host of deep learning models that automate human work, launch constrained actions, and deliver unprecedented value to those in healthcare in-terms of time saved and improved efficiency of people’s workflow.
We imagine that such an interface should require no-training of personnel and should throw answers back at the point of a question. This is not novel, perhaps a sci-Fi concept, an idea that’s demonstrated in films like The Dark Knight, Mission Impossible, and The Fast and Furious. However, sci-fi is no longer sci-fi, but a reality that can be engineered and realized. We believe this allows every individual within the healthcare operations setting to derive information quickly and efficiently. It’s like having an ultimate, all knowing co-pilot inside the pocket of every individual within healthcare powered by latest Large Language Models (LLMs) that are designed to perform in healthcare.
Healthcare being one of the most important pillars for society’s functioning, it’s also one of the most technologically out-dated setups. Coming to Hong Kong I thought much would be automated, but largely public hospitals still use fax machines, Saas installed on computers that has very little room for rapid improvement from appointments being done on phone calls and archaic softwares being used by doctors to deliver patient care, whereas current advances in Large Language Models (LLMs) and general AI can greatly remove bureaucratic process, human work that may stifle swift delivery of patient care. That’s exactly why we’re starting to implement LLMs in private clinics, where doctors can perform various important functions of their day to day lives saving time.
Our frontline health workers, doctors, nurses, and staff need modern tools to deal with modern challenges. These tools can streamline and automate administrative tasks such as appointment scheduling, patient record management, and even analysis of patient medical history. This can allow patients to focus more on patient care rather than paperwork. Besides simplifying the day-to-day tasks for frontline workers, it can greatly reduce human errors, ensuring faster delivery of services, and promote real-time decision making based on data. Additionally handling most processes through autonomous systems can allow for more granular auditing and tracking of operations. Hence, leveraging modern AI technology to augment human capability can enhance operational visibility, efficiency, and patient outcomes.
How can doctors use Clinical Operating Terminal?
Clinical Operating Terminal is our first product that we’re developing. It’s the God’s eye for clinical operations that allows doctors to perform day-to-day functions in the most efficient way possible.
- Drug Search: Doctors can seamlessly search for drugs with the COT. Not only does it provide information about the drug but it also integrates patient-specific information, analyzing the potential effects in relation to the patient’s current condition.
- Analyzing patient condition: The Clinical Operating Terminal allows doctors to analyze a patient’s current medical condition and compare it with the recommended drug or treatment plan. Given GPT-4’s 95% accuracy on United States Medical Licensing Examination (USMLE), it can analyze fairly complex situations. By leveraging the such capabilities, the terminal can generate comprehensive answers and reports and provide insights based on medical examinations and data analysis.
- Professionalizing Notes: Notes can be uploaded in the form of pictures or typed directly into the terminal. Both raw and professionalized versions are saved for future reference, and advanced translation tools are available to ensure effective communication across different languages, enabling easy sharing with concerned individuals.
- Patient Medical History Ingestion: The terminal can ingest patient reports ahead of a scheduled appointment, providing doctors with a summarized overview of the patient’s medical history. It allows the doctor to quickly ask questions and draw connections, facilitating a more efficient consultation process. Doctor doesn’t have to spend time asking questions to the patient, which may lead to mis-understanding, time wasted.
- Chinese Traditional Medicine (CTM) Search: A groundbreaking function of the terminal is the ability to search CTM using natural language. Doctors can retrieve details about specific treatments, herbal remedies, and associated research. Additionally, with the assistance of General AI, doctors can investigate and gain deeper insights into the efficacy and applications of Chinese Traditional Medicine.
- Doctor details and communication: The terminal offers a comprehensive database of doctors’ details, including their professional capabilities, schedules, and contact information. Doctors can access this information easily, enabling them to refer patients to specialists, communicate with colleagues, or seek consultations, all through a single interface.
- Referrals: Doctors can seamlessly generate referrals to other doctors or specialists for further evaluation or treatment. The Clinical Operating Terminal simplifies the referral process, making it more efficient and ensuring a smooth transition of patient care.
- Multilingual: To bridge communication gaps, the terminal supports multi-language user interfaces and translations. A question in any language is translated on the fly and responses are streamed back in the same language, allowing a global usability for doctors in Asian countries, where English is not first language.
The low-code tool for customizing Clinical Operating Terminal:
The low-code is still in its very early stages of development, but its one of the key pillars of the product. It allows for quickly swapping out Language Models with new ones, adding new data sources, and interfaces for interacting with the system. Some of the features we envision are as follows:
- Patient data ingestion
- Connection with patient CMS
- EHR systems
- SQL databases
- PDFs, documents
- OpenFDA and DrugBank for searching drugs
- Traditional Chinese Medicine, and many others.
Every week, there’s new development, new models, new techniques to make systems performant, resilient, accurate, and low-latency. We’re building in the functionality to allow you to place guardrails, define exactly what these LLMs can and cannot see.
- Hydra Chat Interface is our sophisticated chat terminal, which allows streaming responses, translation, and a host of other features and it can be configured according to who is using it.
- WhatsApp: Requisition phone number just by drag and drop. No need to setup and code your own Twilio interface. A simple drag and drop can allow you to have a phone number to start conversations.
- Misc: These are in-loop, but not priority. Telegram, WeChat, Web Chat UI (Popup), as these can allow for a diverse
Every private clinic and healthcare setup is different. It has different operational requirements, data sources, and the need for varying interfaces.
Challenges of deploying Clinical Operating Terminal
Latest advancements in LLMs, and continuous improvements in how data can best be interrogated over documents such as PDFs, emails, files of various formats, data pipelines, and ontologies creates this inevitable opportunity to create natural language interface on top such data, but this comes with a host of problems such as ChatGPT hallucinating, the problems with fine-tuning, data security, the uncontrolled nature of agents, and latency, data visibility, but at Govern Corp our engineering teams are constantly heads down on engineering the systems to address these issues.
Stay tuned for follow up article on how our engineering team is tackling these problems with LLMs and Agents.
The potential to revolutionize the healthcare systems in Hong Kong and worldwide through technology has never been greater. Artificial General Intelligence (AGI) is tipped to permeate every aspect of our lives, necessitating robust construction, deployment, and rigorous testing. This is precisely what we’re dedicated to at Govern Corp. We envision a future where our healthcare infrastructure is seamlessly integrated with cutting-edge technology, promoting efficiency and transparency. Our goal is to leverage the power of software to make significant strides in saving and improving human lives.
Please 📮subscribe here to stay up to date with releases and news about Govern Corp.