Fortune Business Insights™ published the report “Artificial Intelligence Market Forecast, 2022-2029.” As per the report, the global market size was USD 328.34 billion in 2021. AI usage is now deeply integrated into all existing industries - healthcare, aerospace, automotive, tech, food etc. Most frequently it helps to process significant amounts of data (textual as well as visual).

Modelate has developed multiple applications in the area of Artificial Intelligence. For confidentiality reasons, we demonstrate only the products that were already announced or are already on the market.


1. Word Embedding and Entity Recognition for Pharmaceutical Domain

In this project, we have sourced all publicly available abstracts from Medline and Cochrane and processed them in a Hadoop cluster. We trained a neural network to produce domain-specific word embeddings and recognize entities such as “medication”, “disease, etc.,. This allowed us to understand the context of each abstract and search for non-trivial insights. For instance: search for “active substance similar to XXX”, or “if Apixaban is a cure, what is the disease?” and similar.

The working principle of neural networks used for word embeddings

2. Extended version of Entity Recognition tool

This tool uses an improved Entity Recognition library and Machine Learning to understand the context of scientific works automatically. Among all, this tool can recognize study types, population sizes, medications, and general relevance to the search criteria.

An example of the output of the entity recognition

3. AI-based advertisement generator for Uber Amsterdam

This tool automates Banner campaign creation for Uber. Instead of manually creating banners for multiple countries, languages, platforms, and sizes, the tool does this automatically, being given one example of a Banner.

An example of a campaign generated based on one Banner example

4. Road Traffic Analysis System

A customer asked to develop a proof of concept for two cases of automated traffic analysis. In the first case, we had to analyze the load patterns at the intersection. In the second case, we had to detect parking violations automatically.

Our system applied machine learning and computer vision techniques to analyze live video streams. We used an SSD network that performed object detection & segmentation simultaneously and gave up to 60FPS on GTX 1070 GPU.

The advantage of all these cases is that once developed the application improves itself by learning and analyzing bigger and bigger arrays of data. It substitutes the hours of work of specialists - of junior and middle levels.

Back to other articles

Contact us

Value Communication Platform

Ready to get started?
Please give us a brief description of your project, and we will get back to you immediately.

Would you like see VCP in action? Simply enter your data and we will contact you immediately.