LYMPHAblog

Goals

2022-02-10T00:00:00+00:00

Goals/To-do-list

Before any further develpment of lymphaonline or the lymphaapp, the CLI-interpreter has to be improved:

Debugging.
Making every kind of patient-data-evalution work.
Rewrite the regex-recognition and update the lexer function.
Developing example scripts for stroke and myocardial infarction. Making sure that they work without bugs both on the CLI and web project.
Update the pip-version.
Improve how userfriendly the CLI-project is.

This is due to that the CLI-interpreter is the core of every other application.

On communication between healthcare professionals and computers

2022-02-03T00:00:00+00:00

LYMPHA app

Machine learning is usually seen as a future magic solution where data is inserted into a computer. Somehow this will cause the patient to become healthy. Exactly how machine learning can benefit the patient is a difficult question. Some important experimental areas of use are currently the elaboration of new algorithms in the form of flowcharts as well as radiology. In order for machine learning to be used in the day-to-day work of healthcare, a language is needed that is used in the communication from computer to the healthcare professional. Therefore, I also plan to develop LYMPHA in the direction of computers being able to build AI scripts in the language.

References:

https://pubmed.ncbi.nlm.nih.gov/35055347/

LYMPHA App

2022-02-01T00:00:00+00:00

LYMPHA app

The shortage of health care professionals that I have mentioned previously (Patient autonomy and medical algorithms) affects mainly the low-middle income countries. One of the key concerns is severe skills shortages, where technology can be used to help empower lesser tiered workers. In these settings, low-cost phones and tablets, typically Android, are the platforms of choice. Therefore I develop an Android app for that shows what procedure would be appropriate to execute in a certain situation: https://github.com/RickardHultgren/lymphaapp

The app is in a rough alpha stage, but I hope you can get an idea of how the final app should work. The app loads a LYMPHA script. Then each node from the script is displayed as a screen, one at a time. In order to go to the next screen you have to click OK. There is an examle script inside the program with two nodes: crepitation and antibiotics. When you start the app reads the script and then this will be shown on your screen:

The second screen:

Do you want to test the current version? Then just download Kivy: https://kivy.org/#home

At this time, how can LYMPHA be used?

2021-12-15T00:00:00+00:00

Nowadays, medical decisions are often based on medical algorithms. LYMPHA is a programming language designed specifically to handle medical algorithms. The scripts written in the language have the ending “.lympha”. The interpreter for the language is written in Python and runs via prompt. Everything in LYMPHA is represented as constants and variables, but they are called procedures and factors. A common name for these data types is statements. Each statement is represented as a node in the flowchart that the interpreter can produce. The user can choose from different outputs of the interpreter:

As nodes become activated, call an external application via prompt command.
Draw a flowchart through Graphviz where activated nodes are marked.
Print activated nodes’ names

I have also created a web-based variant of the interpreter. Currently, the web version of the interpreter is more advanced. The web variant allows users to create preview-flowcharts and then input patient data that activates the corresponding nodes of the flowchart. Types of nodes that currently the web-based interpreter can handle are procedures, binary factors and binary evaluations. Here is an example for interpartal CTG at term:

CTG. -> CTG_110_160_spm? ; 
CTG. -> CTG_100_or_under_110_spm? ; 
CTG. -> CTG_under_100_spm? ; 
CTG_100_or_under_110_spm? -> corrigate_reversible_causes. ; 
CTG_100_or_under_110_spm? -> consider_stimulation_test. ; 
CTG_under_100_spm? -> corrigate_reversible_causes. ; 
CTG_under_100_spm? -> consider_stimulation_test. ; 
CTG_under_100_spm? -> consider_labour ; 

The script above will produce the following preview-flowchart:

With data from a normal labour the following flowchart will be produced:

Although the development of lympha programs is far from finished, LYMPHA can at least be used as a forework for medical reasoning.

Patient autonomy and medical algorithms

2021-12-09T00:00:00+00:00

Of course, the patient is the chairman of his own “Health Ltd”. But does he also serve as CEO? In the case of a rather acute condition, there are perhaps not many options to choose from. Usually, the physician takes the position of CEO in those situations. But there is an increasing number of chronically ill people. They have to step up and be the CEO since it is up to them to find strategies to live with their condition.

Yet, why not use a machine-learning program that will come up with a suitable solution for most health problems? Machine learning can automate a great deal of medical care. Problems arise where there are no clear, binary answers, e.g. the bus driver, who had to stop working because his hypertension medication caused him to urinate very often. According to the WHO, we have a shortage of health care professionals. But if the workforce increases, what tasks would the staff do? Probably a lot of paperwork. In healthcare today, administrative work is essential, since documentation is the backbone of how it makes money. Machine learning programs could probably handle a lot of that workload. This means that tasks that belong to “soft skilled” work will most likely be more in demand in the future. But a bridge between machine thinking and human thinking is needed. A tool for healthcare professionals that helps them understand the suggestions that the machine learning engine gives. As an example, a doctor explains results from an AI program for a patient. The results suggest a specific path of procedures that should be undertaken. This could be one field where LYMPHA could be used.

The programming language LYMPHA explains medical reasoning. For patients who want to be involved in the decision-making process, the reasoning is crucial. But if the patient thinks that it is up to the doctor to make a proper decision, then the patient is usually not very interested in the reasoning. Future development of LYMPHA will therefore not be focused on applications where the patient is the end-user.

References https://www.who.int/workforcealliance/about/hrh_crisis/en/

A first attempt to write a LYMPHA script for preeclampsia and employing it on LYMPHAonline

2019-06-21T00:00:00+00:00

Introduction

Preeclampsia affects 2 to 8% of all pregnancies and is a leading cause of maternal and perinatal morbidity worldwide. Obesity, which is increasing at an alarming rate, is a risk factor for preeclampsia. Hence preparedness for patients with preeclampsia is of great importance.

One vulnerable population that has a greater risk of preeclampsia is young adolescents. Sometimes the patient might not be aware of the pregnancy. An effective screening strategy could be developed for preeclampsia through the LYMPHA language. Such a tool could even benefit the follow-ups of known pregnancies.

LYMPHAonline is an online interpreter of the LYMPHA language. It is a beta-version, but could be tested as in desicion making support. A LYMPHA script for preeclampsia could be found on: https://github.com/RickardHultgren/LYMPHA-scripts/blob/master/preeclampsia.lympha The script is based on the guidelines of the swedish anestesiologist society.

Instructions

Go to the LYMPHAonline webpage. It takes a while until the entire page is loaded. The screen is loaded when the home-button is orange. In the first box write

https://raw.githubusercontent.com/RickardHultgren/LYMPHA-scripts/master/preeclampsia.lympha

Depending on the situation a healthcare professional could use the script in different ways. If the script should be employed as a screening tool before a visit to the healthcare professional then write in the second box:

screening.

If on the other hand the script should be employed in connection with a visit, then write in the second box:

meeting.

In the third box write how many steps you want to go ahead in the script. If you just want to know what to do next, then write

Then click on Check for varialbe factors and wait. After a while new boxes will appear. There is one box for each parameter that should go into the algorithm of the LYMPHA script.

After clicking Implement variables in script, the script will be executed and a list of action should be shown.