Bibliographic particulars
- Jurisdiction — Republic of Singapore. Certificate issued under Section 35, Patents Act 1994, by the Intellectual Property Office of Singapore.
- Patent number — SG 10202109991Y. Filed 10 September 2021. Granted 6 December 2023.
- Priority — 11 September 2020, Singapore application SG 10202008889X, titled Food Security Exchange System.
- Inventor and proprietor — Zaid Bin Hamzah, founder of FSX.ai.
- Claims — twelve. Claims 1 to 6 are method claims; claims 7 to 12 are corresponding system claims.
The problem addressed
Disasters and threats — macro-environmental events, socio-political upheaval and corporate risk — have direct and indirect impact on food supply, and therefore on food security and resilience. Existing approaches have not dealt with these effectively or efficiently. The invention responds to the need for an early warning system that gives stakeholders in the food ecosystem the guidance to plan for disruption and to take strategic and operational steps to prevent or mitigate it.
The invention in outline
The patent claims both a method and a corresponding system for determining and managing risk in a food supply chain, unified as a single digital platform. The system comprises a processor, memory and a risk determination module, supported by three coupled modules: a natural language processing module, a natural disaster prediction module, and a plant growth monitoring module. A classifier grades the resulting risk.
The claimed sequence
- Receive a preliminary set of data from a plurality of data sources, comprising trigger event data and non-trigger event data.
- Identify the trigger event data from that set using machine learning for natural language processing.
- Obtain weather forecast data — air temperature, humidity, wind speed, air pressure, air density and surface temperature.
- Validate the trigger event data through the natural language processing module.
- Identify the occurrence of natural disaster at a geospatial location indicated by geospatial data.
- Monitor, based on the identified natural disaster, the growth of crops and plants at one or more physical farms at that geospatial location.
- Compare the identified trigger event data against real-time geospatial data and against the weather forecast data.
- Calculate a risk value from that comparison, the risk value including a positive score and a negative score.
- Determine risk by calculating the impact percentile of the risk value; risk is determined where the impact value exceeds a threshold, preset or dynamically set by a user.
- Classify the determined risk into supply-side risk and demand-side risk.
- Alert at least one entity of the food supply chain, based on the determined risk and its classification, via IoT-based sensors.
Dependent claims
- A classifier grades identified trigger events as high, intermediate or low risk, and the risk values of classified events are aggregated to determine overall food supply chain risk.
- Trigger event data comprises keywords relating to pandemic, natural disaster, geopolitical tension and corporate bankruptcy.
- Actions performed on the determined risk include notifying a user above threshold; determining a response plan — including changing suppliers and increasing or decreasing orders; and activating that response plan for each identified trigger event using machine learning algorithms.
- The preliminary data set is obtained by integrating social media data, government data, news data, weather data and company data.
Supporting embodiments in the specification
- Command-and-control centre — an establishment equipped with control and communication instruments, including weather radar, performing monitoring and control across demand and supply, issuing warnings to the distribution network and activating response plans across a supplier network spanning multiple countries.
- IoT sensing — light, soil moisture, temperature and chemical sensors on physical farms, with rainfall, snow, wind direction and speed detectors, plus drone and satellite monitoring.
- Rainfall prediction — a deep neural network model over radar maps at different heights and meteorological sequences, producing time-sequenced rainfall predictions.
- Credibility sub-module — a natural language processing sub-module that scores structured behavioural data against corresponding text data to detect fraudulent, inaccurate or misleading text.
- Financial risk expression — a supply chain risk score derived by factoring an economic impact score against a supply risk score: SCR = EI × SR ÷ 10, each normalised, enabling risks across different production inputs to be compared and ranked, and an ameliorating action taken.
Why it matters commercially
The patent protects the conversion of food security signal into a computed, classified and actioned risk position. It gives FSX.ai a defensible position at the layer where prediction becomes decision — the layer that general-purpose AI tooling does not occupy — and a priority date that predates the current generation of AI platforms. Around it sit three further layers of advantage: the enterprise ontology of the food system, proprietary decision memory, and institutional access across ASEAN and the GCC.
THE GRANTED CLAIMS AS PUBLISHED BY IPOS ARE THE OPERATIVE TEXT AND PREVAIL OVER ANY SUMMARY.
© FSX.AI · ZAID HAMZAH · ALL RIGHTS RESERVED.