DiGi — a startup that predicts earthquakes based on fish behavior

Nurlan Shapanov, city — Bishkek, director of DiGi LLC, @it_digi

How it all started

We are a Kyrgyz company, DiGi, specializing in digitalization and artificial intelligence. Our initiative was inspired by the frequent occurrence of earthquakes in our region, as well as successful cooperation with international organizations, including the UN. The idea to develop a new method for predicting earthquakes based on animal behavior emerged a year ago after a series of tremors. Our country lacks the resources to implement expensive systems, so we are considering alternative solutions. I specifically visited Japan to study their experience in this field.

Our earthquake prediction system based on fish behavior consists of several key components, organically combined to ensure maximum efficiency and accuracy of predictions:

1. Aquarium: specially selected species of fish, sensitive to changes in the electromagnetic background, are kept in the aquarium. The aquarium is equipped with necessary sensors to monitor fish behavior and record changes in their activity, which may indicate an approaching earthquake.

2. Camera: high-sensitivity cameras are installed to observe fish behavior in the aquarium. These cameras can record even slight changes in behavior, which is critical for the accuracy of the analysis.

3. Data transmission system: data collected from sensors and video cameras is transmitted in real time via the data transmission system. This system ensures fast and reliable transfer of information from the aquarium to the server for subsequent analysis.

4. Server: the server processes incoming data, stores it, and prepares it for analysis. This is powerful computing equipment capable of processing large volumes of information without delays.

5. Neural network: the neural network developed by us analyzes data obtained from sensors and video cameras. It is trained to recognize specific patterns of fish behavior that may indicate an approaching earthquake. The neural network can distinguish these patterns from normal behavior or reactions to other stimuli, such as noise or changes in light.

Together, these components form an integrated system capable of accurately analyzing fish behavior and providing early warnings about earthquakes, providing valuable time for response and taking measures to reduce potential damage.

A system based on observing animal behavior to predict earthquakes. The development involves a number of technical and scientific challenges that make the process complex and require deep knowledge in various fields.

The first and one of the most important links is the selection of suitable fish species. Not all species are sensitive to the necessary changes. Our research has shown that certain species of fish react not so much to vibration as to changes in the electromagnetic background, which precede earthquakes. This necessitates careful selection of species capable of perceiving such changes.

Next, the development of a neural network capable of analyzing and interpreting data from sensors is a complex task. The neural network must distinguish normal fish behavior from behavior caused by earthquake precursors. It must be able to ignore false alarms caused by external irritants, such as noise, changes in lighting, or water quality, and accurately determine signals indicating an approaching earthquake.

A huge number of tests are necessary to verify and validate the system. Every aspect, from the selection of fish species to the configuration of the neural network, requires careful testing under various conditions to ensure the reliability and accuracy of predictions.

We have patented our solution, emphasizing its uniqueness and innovative nature. The patent confirms the novelty of our approach to using animals as biosensors for earthquake prediction and highlights the potential of this technology in improving earthquake early warning systems.

Our team began cooperation with a research institute and experts in the field of earth studies. Developers formed the core of our team, actively working on the implementation of the project. 

Target audience

Currently, we are conducting the first tests of our system. The main client is the Ministry of Emergency Situations, but our goal is to make the system accessible and affordable for every resident, including remote rural areas. After successful tests, we will start foreign sales. That we showed the system in March at the UN Headquarters in Vienna is already a big step.

Difficulties

The main difficulty was in choosing the right species of animals sensitive to earthquake precursors in our region. In addition, the neural network turned out to be extremely difficult to develop due to its sensitivity.

Achievements 

We have begun cooperation with the Ministry of Emergency Situations and are preparing for official validation of our system. We also managed to present the first results of the project at a UN event in Vienna. At the moment, anyone can buy our aquarium, although we still have to complete the training of the neural network. The accuracy of the system is already impressive.

Plans

Our main task is to conduct tests, tests, and more tests to improve and confirm the effectiveness of our system. We hope to deploy the system already this year.