Dear MongoDB Community:
My name is Prasad Reddy. I am a Internal Medicine physician by training and managed a group of physicians in that regard. I also have some understanding of RDBMS and overview of various types of big data technologies.
Currently, the world is faced with Covid Pandemic and current Electronic Medical Record systems ( EMR) are not ideal to use in this setting. They provide much richer data, but if the patients are being treated in Central Park of NY, fancy EMR is useless. You can not set up servers, network infrastructure etc in Central Park quickly.
** Cloud based solution will solve this.
To minimize the risk to healthcare workers ( CNAs, RNs and MDs), we need to have as much of the healthcare to be done on telemedicine mode. EMRs do provide some remote access to physicians, but nurses unfortunately still need to be on the battle ground. Some of the functions that the doctors and nurse do for example history taking of the patient, can be done by minimal training by others. A call center employee and recently furloughed airlines employees can quickly be trained to take medical questionnaires from patients to relieve healthcare workers to focus on ICU patients where more resources are needed. Low acuity patients that are asked to stay at home have no real monitoring. Collecting a daily medical " Review of Systems" by call center employees can pick up potential hospital admissions ahead of time. Current EMRs can not do that.
** A could based EMR can do that.
While there are some cloud based EMRs, none of them are designed for billion patients in one database with million concurrent users. This requires writing the project from ground up.
I am working on EMR back end data components. In this regard, while keeping very high scale as priority, I have converted most of the data elements of a medical record into template driven ( few clicks instead of note typing ) elements. While my model may loose some data richness, it helps in scale. I am using only two types of document templates ( One short version and one long version ) with MongoDB documents. These two documents types ( Long and Short ) will address about 4000 types of data from Vitals signs, labs to medications and diagnosis.
In my design each MongoDB document would have EVAT Plus some metadata fields.
By converting the medical data into EAVT ( Entity, Attribute, Value, Time) format and then adding some metadata fields in the document itself, this data can be moved to RDBMS or Columnar Databases for faster query across billions of records. Amazon EMR could be good solution, or may be MongoDB can handle this with more clusters. I am not sure on this.
Also this EAVT + Metadata approach allows the database to be converted to other regions of the world and languages very easy to achieve.
Adding a cheap smart watch to each patient can add the Heart Rate in streaming fashion and that can provide a better insight to patient condition than regular vital signs. By developing AI algorithms, patient deterioration can be predicted much quicker and helps to risk classify the patients. This would be much better than the existing clinical prediction scores ( APACHI II, SOAP score, SAP II score ) None of the existing clinical prediction scores are based on streaming vital signs. This step alone would save hundreds of thousands of hours nursing time, while providing better insight to patient’s condition.
This is the top level summary of the project. I am working on this project with an intention to help community, but noticed that working alone would take much longer time to develop a working product. I am wondering if anyone of you can lend me a hand in this project. This product would be fully open source and all the technology used has to be opensource as well.
I have looked at MongoDB with Atlas as best solution for the project hosted on Gov Cloud as best way to implement. I hope that this project would be taken up by the governments once it is developed to the extent that they see the value as well as trust the design principles.
At this stage, I am hoping to get help on setting up a cluster on AWS as well as identity the best middle layer and front end frame work that can be added to the MongoDB/Atlas. Once done, I have few ideas on wire frames to implement.
My LinkedIn profile is
Any help in this endeavor is greatly appreciated. Please connect with me on LinkedIn and I am happy to connect right away. If you can reply to this thread or reply to LinkedIn, I can connect quickly.
If this topic is better suited in some other section, please do the needful.
Prasad Reddy, MD
Salt Lake City, UT, USA
April 4th, 2020