Market Context — Why This Technology, Why Now

The increasing complexity of global supply chains, financial markets, and healthcare demands more sophisticated predictive analytics. Traditional AI models, optimized for single tasks, are proving insufficient and costly to scale. This creates a critical need for integrated AI solutions that can process diverse data streams and provide holistic insights, driving efficiency and reducing operational overhead across industries.

Key Competitive Advantages
01

Enhances Efficiency by ~20% with Integrated Multitask Learning

02

Achieves High-Accuracy Prediction for Complex Time-Series Data with Recurrent Neural Networks

03

Provides Market Entry Advantage Due to Limited Prior Art

Market Opportunity
🏭 Smart Factory
$10B–$15B globally (AI est.)
Growing demand to maximize production efficiency by integrally learning multiple tasks such as quality control, anomaly detection, and predictive maintenance on production lines.
Industrial automation solution providers Manufacturing equipment OEMs Large-scale factory operators
💰 Financial & Insurance
$8B–$12B globally (AI est.)
High-precision predictive models based on complex data, including credit assessment, fraud detection, market forecasting, and customer behavior analysis, are key competitive differentiators.
Fintech solution developers Major banking institutions Insurance technology providers
🏥 Healthcare & Medical
$4B–$6B globally (AI est.)
Expected application in AI diagnostic support for comprehensively evaluating disease risk and predicting treatment efficacy from multiple biological data points and test results.
Medical imaging AI developers Pharmaceutical R&D firms Digital health platform providers
IP Defensibility — Why Competitors Can't Replicate This
What This Patent Covers

This patent protects an information processing apparatus and method across a broad scope, supported by 7 claims. The successful overcoming of an office action demonstrates the robustness and stability of the claims against future invalidation challenges, ensuring strong defensive capabilities.

Competitive White Space

This patent focuses on the core multitask learning algorithm using RNNs. It does not explicitly cover specific hardware implementations for AI accelerators or novel data preprocessing techniques, offering white space for licensees to develop proprietary solutions in these areas.

Economic Impact
~$105K/year estimated AI development and operational cost savings per facility (est.).
estimated ROI · USD · AI analysis
ROI Calculation Logic

This technology could integrate and reduce development and operational costs for multiple single-task AI models. For example, replacing three individual models with one multitask model could reduce development effort by 15 person-months. Assuming a personnel cost of ~$5,350/month (AI est.), this could result in an annual development cost reduction of ~$80K (AI est.). Additionally, operational costs could be reduced by ~$25K (AI est.) annually, totaling an estimated annual cost reduction of ~$105K (AI est.).

Speed to Market
6× faster than in-house development
This technology's algorithm is established as a research outcome from Kyushu Institute of Technology, with a robust technical foundation already in place. Developing a multitask learning model using recurrent neural networks from scratch would require securing specialized personnel and several years of R&D. Licensing this patent could significantly reduce that time and cost. Given the willingness to license, rapid integration into existing AI infrastructure and data platforms is anticipated.
Competitive Positioning

X: Predictive Model Versatility
Y: Complex Task Processing Capability

Business Models & Applications
☁️ SaaS Predictive Analytics Platform
Offer a cloud-based predictive analytics platform powered by this technology. Customers can integrate data via API to leverage high-accuracy predictive models through multitask learning.
🧩 AI Engine Embedding License
Provide licenses to embed this technology's AI engine into existing business systems or products. Licensees can enhance their solutions' value and strengthen market competitiveness.
🔬 Industry-Specific Solution Development
A business model for co-developing and providing specialized multitask learning solutions for specific industries, such as smart factories or financial risk management.
Adjacent Application Opportunities
👵 Elder Care & Monitoring
Integrated Vital Sign Prediction System
This technology could integrally learn data from multiple sensors (heart rate, respiration, activity, sleep patterns) to predict sudden health changes, fall risks, or signs of cognitive decline early. This could enable more personalized monitoring services and reduce the burden on care providers.
🏭 Smart Factory
Integrated Manufacturing Process Optimization AI
By simultaneously learning sensor data (temperature, pressure, vibration, images) from each manufacturing step and final product quality data, this technology could integrate anomaly detection, quality prediction, yield improvement, and predictive maintenance for equipment. This is expected to maximize production efficiency and minimize defect rates.
🏥 Medical Diagnosis Support
Complex Disease Risk Prediction AI
This technology could integrally analyze and learn multiple types of medical data, such as patient electronic health records, imaging diagnostics, genetic information, and interview results, to comprehensively predict the onset risk or progression of specific diseases. It is expected to serve as a valuable diagnostic support tool for physicians.
Integration Roadmap — Estimated 17-Month Deployment
Phase 1: Technology Evaluation & Requirements Definition
Duration: 3 months
Evaluate the compatibility of this technology's core algorithms with existing systems and define requirements based on the licensee's specific business challenges.
Phase 2: Prototype Development & Validation
Duration: 5 months
Develop a prototype incorporating this technology based on defined requirements. Conduct validation using real data to assess performance and identify areas for improvement.
Phase 3: Production System Deployment & Optimization
Duration: 9 months
Build and deploy the production system based on prototype validation results. Optimize performance through continuous data learning and model refinement.
Technical Feasibility
This technology primarily focuses on creating learning models applicable to recurrent neural networks and can be implemented as software on major AI frameworks. The 'reception unit' and 'processing unit' mentioned in the patent claims can be easily built on existing data processing pipelines or cloud infrastructure, eliminating the need for large-scale new equipment investment. Utilizing existing computational resources and data integration via APIs is expected to enable relatively low-cost and rapid system integration.
Success Scenario
Upon adoption, this technology could allow companies to integrate multiple individually operated predictive models into a single multitask learning model. This is estimated to reduce model management effort by approximately 30% and optimize resource utilization. Furthermore, by leveraging information from multiple tasks, prediction accuracy could improve by an average of 15%, enabling faster and more precise decision-making.
Patent Record
APPLICATION NO.
特願2021-007423
REGISTRATION NO.
7551114
FILING DATE
2021/01/20
GRANT DATE
2024/09/06
EXPIRATION DATE
2041/01/20
PATENT HOLDER
国立大学法人九州工業大学
Examination History
2021年02月15日
手続補正書（自発・内容）
2023年09月22日
出願審査請求書
2024年04月23日
拒絶理由通知書
2024年05月24日
手続補正書（自発・内容）
2024年05月24日
意見書
2024年08月13日
特許査定