You are absolutely right. What we are going to do about it? I do propose to write down a methodical paper that will cover all of our areas of expertise. What is your opinion? It is meant to both of you Dr. Bykova and Dr. Cukic.

Sure. Although it is a very complicated process. Process-based/mechanistic species distribution models (SDMs) that incorporate various climate change scenarios and mechanisms controlling species distribution are able to estimate future biodiversity responses. Although these models are still under the development, the existing results are very promissing.

following

Yes, it can be used in medicine to predict development and onset of diseases. I do personally use AI techniques to predict life-threatening arrhythmias. There is a software that predicts arrhythmias about one hour prior to their onset on a rabbit model.

The prediction has 93% sensitivity and specificity and is used to predict medical drug-induced Torsades de Pointes arrhythmias, a kind of deadly ventricular arrhythmias, that are killing a lot of people who take those potentially deadly drugs for diseases that are usually not related to the heart.

Even more big problem is that people dying on TdPs has very often no sign of any heart disease. TdPs here recently classified as a silent killer. This method of approach of prediction of TdPs can be extended to all ventricular arrhythmias.

The principle of the method is to mathematically process the original ECG curve and apply AI to distill most important features and markers that are having high predictive power.

The discussed approach was indirectly proved by another group working on human ECGs that were taken at ICU, they got specificity and sensitivity of 89% by using deep learning techniques.

Personally, I do see that in the future medicine will gain the ability, thanks to mathematicians and computer scientists, to be predictive, proactive, and personalized.

Methods that I am developing with some other colleagues, independently working in other countries, will lead to a complete change of our understanding of medicine. AI will lead medicine from the age of reductionism to the era of systemic approaches.

There is existing the most important way of predicting the future development of the observed phenomena: mathematical and computational modeling.

I would like to turn your attention to a relatively new, less known approach, that has a great potential to predict the future in the coming decades: complex systems with their self-organizing and emergent properties.

Those who would like to know more about the subject can read the introductory review paper prepared for this and similar occasions:

https://www.researchgate.net/publication/330546521_Complex_Systems_and_Their_Use_in_Medicine_Concepts_Methods_and_Bio-Medical_Applications?_sg=gN_UY91x92OoRArjG5IJa87d0AjZLnxsmZD7pIc7F1OPklJpGv7VoeIusN2PJ88Vt6rU6bBNLo8-BA.zH-QCAK4Z9DvRtDLNC-WG6bmGmPrHJ4DwoFojotr3_YOhw4nrJoWrnvhPJeJa6DHv95_KUqp6jFw9yVIoeh16Q&_sgd%5Bnc%5D=0&_sgd%5Bncwor%5D=0

Yes, prediction in medicine is possible. But for that you need to apply wery sophisticated techniques to test tge reluability of your machine learning model, and collect a lit of data. What you need is a good generalisation so you have reliable prediction on unseen test data.

Completely agree with you, dear colleague, Milena Cukic. That is why we desperately need well defined, rock solid, unmovable, biomedical data which are intact, not manipulated with.

We need those data as open source because only in this way is research reproducible, responsible, unfalsifiable, and safe.

My current research on the prediction of arrhythmias is a children toy compares to what we are speaking about.

It would be necessary to force the whole system towards such setups. We are still at the stage of small research groups collecting their little data and doing great predictions with nit so great reliability due to lack of sufficiently large data databases.

I would be more than happy to put my software on a database of thousands of cases of arrhythmias (ventricular tachycardias are sufficient now) and the whole thing can proceed forward quite quickly.

How can we make any conclusions when we have only tens or one hundred cases to learn and test our software on?

A paraphrase of an old sentence can be: "Give me a stable and reliable database and will move the whole medicine!"

What is interesting is that although we work in completely different research fields, the major problem of modelling at this point is the luck of solid emperical data.

You are absolutely right. What we are going to do about it? I do propose to write down a methodical paper that will cover all of our areas of expertise. What is your opinion? It is meant to both of you Dr. Bykova and Dr. Cukic.

Good idea. Count me in!

Good news :-) We three can write down a methodical paper about what should be done in the collection of data and what to check before someone starts to work on research. And how to proceed.

I saw a lot of good AI and ML papers where authors, usually young researchers, were lacking the information about those issues and it decreased the quality of otherwise high-quality work.

It can be used even by those who will create those publicly available medical data databases.

I agree that this kind of research is very important especially to people in neuroscience.

I hope what we work can be of help to other colleagues.

Yes, you can use artificial intelligence (AI) for prediction. It requires availability of sufficient data (input, output) to design prediction model. There are many methods for training the data to develop and test the model. Neural network is one of AI tools for developing forecasting models. A software package that contains such tools is required (MATLAB, Python or latest versions of SPSS).

This a quick introduction may contribute to reply the question.

In view of the above, it is also important to include in the above discussion also the following issue:

What elements of information technology should an analytical and database system be used to predict the continuation of trends for multifaceted processes?

The analysis of predictions and the continuation of specific processes could, for example, relate to climatic, market, cosmic, geological, natural and economic processes, etc. Are the basic components of this type of system should be the following components of computerized advanced information processing technologies typical of the current technological revolution, known as Industry 4.0.?: - Big Data database system, - Business Intelligence analytical platform, - data processing in the cloud, - artificial intelligence. In addition, if the system should have convenient access from the Internet, these types of solutions can be made possible through the Internet of Things technology. In this way, access to the system or some of its modules could be implemented, for example, from smartphones, tablets, laptops and other mobile devices.

Do you agree with me on the above matter?

In the context of the above issues, I am asking you the following question: What elements of information technology should an analytical and database system be used to predict the continuation of trends for multifaceted processes?

Please reply

I invite you to the discussion

Thank you very much

Best wishes