Inventale

Client Company 

A sports-betting operator based in the UK

Solution Overview 

SPORTEX is a proof-of-concept (POC) project designed to enhance gaming and sports betting experiences for platform operators and odds-making teams. Initially focused on football match analysis, the platform provides near real-time positional and relative statistics through video recognition. It offers advanced functionality, including performance predictions based on historical data.

Basic Version Includes:

• Player Recognition: Identifying team, player number, appearance, position, and goal involvement, as well as measuring player pace and speed.

• Player Actions and Outputs: Capturing tackles, tackle breaks, fouls, turnovers, and other relevant player actions.

• Set Play Actions: Analyzing throw-ins, corners, free kicks, goal kicks, penalties, and their locations.

• Set Play Results: Evaluating the outcome of set plays, including initiation, action result, success, and failure.

• Ball Movements and Actions: Tracking ball movements, positions, action types (shots, passes), and success rates.

Advanced Capabilities Include:

• Player and Team Performance: Comparing current performance to historical data, considering player pace, activity, team position, and ball possession.

• Player Actions with the Ball: Analyzing player performance during ball interactions, taking historical data into account.

• Potential Injury Identification: Detecting sudden changes in player pace or performance that may indicate potential injuries.

• Performance Stats Relative to Grounds and Pitch Conditions: Assessing performance statistics in relation to the playing field (home or new field) and pitch conditions (weather, time of day).

• Performance Predictions: Utilizing historical data and making real-time adjustments to generate performance predictions for players and teams.

Technology Background 

The solution relies on neural networks and machine learning algorithms, incorporating the following technologies:

Video Processing: Utilizing the open-cv library for video processing, following a classical pipeline approach. Object Detection: Implementing a fast, open-source solution based on the dlib C++ Library for accurate object detection. Person Recognition: Employing machine learning models based on the TensorFlow framework for reliable person recognition. Unique Counting: Utilizing a proprietary hashing algorithm for future re-identification of individuals. Demographics: Leveraging services that provide data on BSSID and its location, such as GPS data, for demographics checking.

Results and Current Status 

The POC has demonstrated success, and the applied technologies and algorithms are now being tested for horse-racing analytics and forecasting. With the growing trend towards automated video statistics generation, reporting, and data mining, this technology holds significant potential for various fields beyond sports betting.

Client Overview

NexusMentis is a testament to overcoming challenges. This project is supported by a private investor who personally experienced the difficulties faced by modern psychiatry in finding timely and effective treatments due to limited data and fragmented knowledge. Thus, the idea was born to incorporate artificial intelligence to speed up identifying the best medication and facilitate the exchange of experiences among doctors, all with the ultimate goal of enhancing people's well-being.

Challenge 

The client aimed to:

• Provide a solution that utilizes AI to offer recommendations for suitable treatments, reducing inadequate prescriptions and enabling knowledge sharing among doctors.

• Develop an expert system capable of analyzing millions of combinations and suggesting the most appropriate medication from a comprehensive knowledge base.

• Automate routine tasks for doctors, such as patient surveys, visit journals, and appointment scheduling.

Solution Overview 

NexusMentis is an AI-based platform designed for psychiatrists, providing personalized recommendations for each patient.

The system analyzes anonymized aggregated data, including past prescription histories, medication compatibilities, daily patient surveys, mental states over a period, and other patient narratives. Based on this analysis, it advises doctors on the optimal choice of medication. The platform consists of a desktop version for doctors and a mobile application for patients.

How It Works:

• An integrated interface for doctors manages patient records and schedules and offers advanced prescription recommendations and possible diagnostics.

• The algorithms learn from patient data entries and doctor decisions, generating recommendations based on:

  • Anonymized medical records of the doctor's patients.

  • Case histories of hospital patients with psychiatry-related diagnoses (provided by an independent data collector).

  • Medication databases.

  • Stahl's Essential Psychopharmacology, and more.

• A mobile app for patients provides daily surveys on the patient's condition, sends reminder notifications, and offers recommendations for meditation, breathing exercises, etc.

• An app integrated with the doctor's interface monitors the patient's well-being and response to medication.

• The app securely transmits data to the interface through an encrypted gateway and stores it in the cloud. Doctors utilize NexusMentis recommendations when making final treatment decisions.

Technology Background 

The system is built upon neural networks, machine learning, and AI algorithms. The core of the NexusMentis recommendation engine relies on factorization machines (FM), deep machine learning algorithms, and big data processing methodologies to generate meaningful recommendations.

Key Features of the Recommendation Engine:

• Collaborative Filtering: Determines user profiles and recommends medications based on patients with similar profiles who achieved the best results.

• Factorization Machines: Searches for correlations between user attributes and products using substantial sparse data.

• Regression Algorithms: Utilizes techniques like Bayes/DecisionTree/Support Vector Machine to predict the probability of a specific effect based on a chosen user.

Results and Current Status 

NexusMentis is a unique start-up in the pre-seed phase, operating as a decision-support recommendation system for doctors through its Smart Pharmacy and diagnostic tools.

Accomplishments to date:

• Inventale has developed a recommendation engine that matches diagnoses with medicines and suggests the most suitable medications.

• The recommendation engine has undergone training and testing using a comprehensive database.

• The beta version of the system was presented to the UAE Ministry of Health & Prevention (MoHAP) and received highly positive feedback, expert opinions, and valuable advice.

• Based on MoHAP's recommendations, Inventale's team enhanced the NexusMentis interface for doctors by incorporating several new features.

Client Company:

B2Connect is a telecom company based in the UAE that has created the world's first community platform for hand-picked products and services. It aims to connect and improve the well-being of residents in staff accommodations and workers' villages across the GCC.

Challenge:

In today's world, Wi-Fi has become essential for any business. B2Connect was founded on the belief that Wi-Fi connectivity should be fast and reliable, no matter where or when you need it.

B2Connect's goals were to:

• Provide Wi-Fi connectivity throughout large estates.

• Enhance the efficiency of internet kiosks, addressing issues with client payment records and management.

• Effectively resolve network management problems in staff accommodations.

• Develop a mobile application for managing Wi-Fi packages and payments.

• Consolidate disparate information about tariffs, clients, and payments into a single system.

• Promote B2Connect's e-shop through the developed mobile app.

Solution Description:

Inventale has created a user-friendly mobile app for B2 Connect with two modules: Smart Wi-Fi and Easy Shopping. Users can pay for Wi-Fi subscriptions or purchase the latest gadgets and accessories with just a few clicks.

The platform relies on Wi-Fi and user data analytics, providing the following capabilities:

• Consolidating client data into a unified database containing all the necessary information about users, payments, tariffs, devices, etc.

• Detecting and continuously monitoring anomalies and fraudulent activities, such as multiple user entries under the same account.

• Offering real-time statistics and analytics about platform users, including their gender, location, payments, etc.

• Integrating a notification system to send timely reminders about billing plan expiration and customizable notification settings.

Technology Background:

The backend infrastructure is fully deployed in the AWS cloud. By utilizing Infrastructure-as-a-Code patterns, the system achieves horizontal scaling and enables the rapid deployment of multiple stands, such as staging and QA environments.

The microservice architecture allows separate deployment and maintenance of different product components, such as the backend for mobile applications, internet kiosks, management UI, management API, and data synchronization applications.

The Android client and the backend have been developed using the modern Kotlin language, incorporating cutting-edge frameworks, libraries, and technologies like document recognition.

Results and Current Status:

Inventale's team has successfully met all of B2 Connect's requirements, delivering the following solutions:

• A mobile application that empowers users to manage Wi-Fi packages and access B2 Connect's e-shop. Additionally, a news integrator was embedded at the client's request.

• The solution provides the client complete control over their users and business processes, serving as a monitoring and management portal.

• The system integrates with payment, e-commerce, and user management services, consolidating everything into one platform.

• Internet kiosks are synchronized with the client's inventory system, resolving user payment management and control issues.

A portal for analyzing data on users' movements, building user segments based on their interests, and providing analytics on places with users' concentrations and their daily journeys, customers' loyalty, trends, variances, and more that helped Builders to choose the best location for their new store. 

Builders is a South African chain of DIY stores from Walmart Group. In 2019, the company decided to open another Builders Express near Johannesburg, and Inventale was their technical partner that assisted Builders in finding the best spot based on users' geodata analytics.

Due to the developed algorithms, it took Inventale a week to evaluate Cresta Shopping Centre as a primary option and assess other possible variants. During this time, Inventale identified Builders' main competitors, counted the actual number of visitors to all nearby stores, found out the areas they live, how frequently they go shopping, and how far they are ready to go for it and assessed the risks of cannibalization for other Builders stores. 

As a result, Cresta Shopping Centre was chosen as a new place for the store. After this, the client continued to use Inventale's analytics portal for more than a year on an ongoing basis to: 

• accurately segment the offline audience by their interests (e.g., car enthusiasts, young parents, new buildings residents, etc.);
• analyze the routes of the target audience;
• analyze routes of and segment competitors' buyers;
• find suitable locations for advertising billboards to target competitors' audiences.

A portal that allows one to analyze the history of users’ movements, build user segments, and provide all kinds of analytics like places with users' concentrations, customers’ loyalty, trends, variances and many more. 

Backlite Media is the leading outdoor advertising company in Dubai that provides premium billboards and out of home advertising. Backlite Media was looking for a software solution to analyze mobile, Wi-Fi and GPS-data on the users passing by their advertising billboards for two business purposes: 

• choose the best locations for their advertising stands and adjust advertising policy based on users' interests and demographics; 
• collect and analyze statistics on the ads view rates.

The developed portal enabled the client to segment users based on their routes and digital fingerprint, points of interest, reconstruct their gender and age based on look-alike modeling, and provide all types of analytics like places with users' concentrations, customers' loyalty, trends, variances, etc. 

The Portal uses various technologies for achieving its business goals:

• AI tools to get a user's routes between venues
• API/AWS/FTP integration for audience segments export to various DDP systems such as Google AdManager, AdForm, Lotame, etc.
• Competitors’ analytics & visualization in a high-performance web application: number of their visitors/customers; which our customers attend competitors' schools/shops; other places of interest of our customers. 

And Big Data processing cloud for: 

• building audience segments based on user interests and places they visit, like "young parents," "style&fashion lovers," "fine diners." Interest profiles of our visitors/customers;
• customer loyalty analytics – how often do the customers come back to a shopping center/car dealer, the number of new/loyal customers 
• identification of customer resident/working areas.

Inventale worked with Dragoman on the brand audit service for Majid AL Futtaim Group. The project's goal was to provide senior management with quantifiable year-on-year statistics of the organization's corporate brand compliance. Inventale and Dragoman developed an online brand management portal to house all corporate brand guidelines and assets essential to the rollout and management of the client. 

Majid AL Futtaim is the leading shopping mall, hotel, leisure, and retail company in the Middle East, Africa, and Asia, covering 15 international markets and employing 43,000+ people. When in 2014, Majid AL Futtaim consolidated its brand strategy under one master brand, it required a brand affinity platform for constant brand compliance monitoring. 

Inventale and our partner Dragoman developed a portal for brand audit and analytics for all Majid AL Futtaim properties. The platform included three types of user roles for solving different tasks: audit assignment, audit results check, and comprehensive analytics. It offered very flexible tasks and templates building functionality, export, and sharing possibilities. At that, the system's primary goal was to have everything in one place and enable users to set standards, control, monitor, and perform all types of analytics.      

Verification of real users' location by Wi-Fi, Bluetooth, mobile data networks, GPS (for some devices - humidity/pressure/gravity sensors), and revealing the attempts to conceal their actual location. 

The client is a US-based network of online casinos that, according to US legislation, can provide access to gambling games only in specific states and in dedicated game zones.

The exponential growth of technology enables users to find more and more opportunities to falsify their location. Hence, more than standard approaches to determining the actual user's position is required. Checking the location by GPS is unreliable as the GPS data are easy to work around.  

Inventale's client asked for implementing a mobile SDK that would spot the actual user location by their geo-data (Wi-Fi, Bluetooth, mobile data, GPS) and reveal the attempts to conceal actual location.

The delivered solution: 

• checks location according to cellular, and Wi-Fi data (for Android, the used Wi-Fi network and all available ones are analyzed) and system location service;  
• examines data on the device (absence or fakeness);
• detects an IP change: the IP address is checked every minute, and in case of any changes, the device sends a push notification about IP change;  
• validates user application;  
• runs a check via services that provide data on BSSID and its location (eg. GPS data).

A system for a broadcasting company that predicts movies' and TV series' popularity at the planning stage and enables one to choose the ones with the highest potential. 

A large media channel producing TV and video content was looking for a software solution to enable it to estimate a TV series' popularity at the stage of its creation. The solution was to help choose between several plot ideas and/or tune a specific one. 

Inventale's tremendous experience in machine learning, working with Big Data, and delivering forecasting and recommendations systems helped immensely with this project. 

Hundreds of parameters were analyzed, such as:

• genres and settings;
• cast and film directors;
• the correlation between the year of release and the film or series rating;
• the temporal genre popularity;
• actors, screenwriters, and directors ratings, depending on the genre;
• socio-demographic perception, etc.

The forecast also considered the following factors:

• seasonality; 
• the competition of major projects and TV shows between channels;
• competition with significant scheduled events (like Super Bowl or the World Cup), and others.  

The recommendations done by the system were fully considered by the client when choosing and adjusting the best TV series' idea.