Field Testing and Validation

Prepared by: Apiculture Technology International, LLC for Institutional Partners, Funding Agencies, and Research Collaborators Date: June 28, 2025

1. Objective

The Field Validation Program aims to establish real-world performance benchmarks for a unified chemical and environmental sensing system designed to detect early indicators of hive stress, disease onset, and agrochemical exposure. While laboratory data and simulated deployments have confirmed the functionality of the underlying Chemical Sensing Module (CSM), field validation is required to assess operational reliability, inference accuracy, and user integration under diverse environmental and biological conditions.

This multi-site program spans deployments in Austin, Texas and central and northern Thailand, enabling cross-continental validation across both temperate and tropical ecosystems. These locations were selected to capture year-round colony activity, regional pesticide practices, and distinct hive architectures (Langstroth for Apis; traditional boxes for Melipona/Trigona). Field validation will proceed from Q3 2025 through Q4 2026 in four planned phases—baseline, active monitoring, diagnostic comparison, and final synthesis.

The program evaluates three core sensing devices, all based on the proprietary CSM/CSMA platform:

• HiveShield™ – Internal sensor for Apis mellifera and Apis cerana colonies
• MeliponaShield™ – External sensing variant for stingless bee species (Melipona, Trigona)
• TransitShield™ – Internal sensor for Apis mellifera and Apis cerana colonies while in transit

Field validation is essential not only to verify the technical resilience of these systems in beekeeping and agricultural contexts, but also to generate the high-quality labeled datasets required for refining AI-based classification models used in early detection, response, and forecasting tools.

2. Field Validation Goals

2.1. Technical Validation

1. Measure the accuracy and repeatability of VOC/VSC detection under live deployment
2. Evaluate algorithmic classification of stress- and contamination-related gas profiles
3. Confirm environmental resilience of enclosures (temperature, humidity, contamination)
4. Validate dual-mode performance (Baseline Trend vs. Inference Mode)

2.2. Biological & Agricultural Relevance

1. Correlate gas signal shifts with observed bee health or mortality
2. Confirm early detection of brood decay, hive stress, or disease onset
3. Detect and classify agrochemical exposure events in organic/perimeter zones
4. Assess real-time vs. retrospective detection value for apiary management

2.3. Operational Metrics

1. Measure battery life under typical use conditions
2. Assess BLE setup and LoRa transmission success rates
3. Track baseline stability and signal drift over time
4. Determine adaptability across diverse hive architectures and field placements

2.4. Epizootiological & Research Utility

1. Generate traceable, geotagged chemical exposure logs
2. Enable long-term data collection for AI model training
3. Support early warning systems and risk forecasting tools
4. Provide anonymized datasets for research collaboration and publication

3. Deployment Framework

3.1. Hive and Site Acquisition

1. 100 × Apis hives (Langstroth, disease-free, geographically distributed)
2. 25 × Melipona/Trigona hives (sourced in Southeast Asia, specifically Ratchaburi, Thailand)
3. 15–25 TransitShield™ during transportation of hives used in the pollination industry

3.2. Installation Strategy

1. HiveShield™ and TransitShield™ units suspended between brood frames (or embedded into frame top bar)
2. MeliponaShield™ devices mounted externally or top-inserted using diffusion ports

4. Test Phases

5. Data Collection & Analysis

Sampling Frequency

1. VOC/VSC: 1–2× nightly
2. Temperature/humidity: every 30 minutes
3. LoRa transmission: once per measurement set

Data Types Collected

1. Raw resistance data (per heater cycle)
2. Feature vectors for inference models
3. Alert classifications (trend, compound group)
4. Battery level, signal strength, uptime logs
5. Metadata: Hive ID, location, timestamp, thresholds

Analysis Tools

1. ROC curves and confusion matrices for classifier performance
2. Drift analysis for sensor stability
3. Geographic distribution heatmaps for VOC anomalies
4. Event clustering over time to detect systemic patterns

6. Collaboration Roles

TBD

7. Deliverables

1. Secure online dashboard for internal + optional stakeholder access
2. Multi-phase technical validation report
3. Peer-reviewed publication(s) or whitepapers
4. Machine-readable dataset export (VOC trends, events, environmental data)
5. Annotated training set for AI inference refinement
6. Deployment photo log and field protocol summary

8. Success Metrics

TBD

9. Budget Considerations

1. Hive procurement and setup (Apis and Melipona)
2. Sensor fabrication (HiveShield™, MeliponaShield™, TransitShield™ units)
3. Field travel and on-site support
4. Agricultural and microbial lab analysis services
5. LoRa network licensing (where applicable)
6. Web hosting and dashboard maintenance
7. Publication and open-data preparation

10. Ethical and Environmental Compliance

1. No chemicals introduced for testing purposes
2. Passive, non-invasive sampling only
3. Sensors do not interfere with colony behavior or hive thermoregulation
4. All data anonymized for public datasets
5. Consent obtained for all partner site deployments

Integration with AESRS Platform

All data collection efforts will be supported by our already developed Apiculture Epizootiological Surveillance and Response System (AESRS) — a secure, multi-user, web-accessible platform for real-time data visualization, historical analysis, and alert management. Each deployed device transmits GNSS-tagged environmental and/or acoustic data to the AESRS database via BLE or LoRa uplink. The system enables centralized monitoring of hive health trends, VOC anomalies, pest alerts, and chemical exposure events across geographically distributed sites. The platform supports both field research and regulatory use cases, with role-based access, multi-tiered data filtering, and export functionality for validated datasets.

Mr. McIntosh – Principal Investigator

Mr. McIntosh will serve as the Principal Investigator. As the technology Inventor and Lead Product Developer, he will provide direct technical oversight, integration of hardware and firmware systems, coordination of device deployment, and authorship of final technical reports. Mr. McIntosh will oversee both the U.S.-based and international components of the project, ensuring consistency in device configuration, data integrity, and validation protocols.

Mr. Willoughby – Co-Investigator (Thailand Field Research Lead)

Mr. Willoughby will serve as the Co-Investigator and Field Research Lead for Southeast Asia. He will manage sensor deployments in Apis, Melipona and Trigona hives, coordinate with local institutional partners, oversee data collection, and support field-level evaluation of system alerts. He will work closely with Mr. McIntosh and our university partners at King Mongkut’s University of Technology Thonburi (KMUTT) and Chiang Mai University (CMU) to ensure procedural alignment across regions.

Project Schedule Summary

The field validation program will be carried out over a period of approximately 15 months, beginning with a 1-month acquisition and preparation phase during which hives, field sites, and sensor hardware will be secured and configured. This will be followed by:

Phase I (1 month): Baseline calibration and system deployment

Phase II (6–8 months): Active monitoring and data collection under real-world conditions

Phase III (2–3 months): Diagnostic comparison, behavioral analysis, and ground-truth verification

Phase IV (1 month): Final data synthesis, AI model refinement, and report generation

All phases are supported by the AESRS data infrastructure, enabling synchronized data capture, remote monitoring, and performance logging across both U.S. and Thailand-based sites.

Testing for HiveShield™, MeliponaShield™, and TransitShield™ will be conducted across strategically selected sites to maximize environmental diversity and data relevance. HiveShield™ units will be deployed in 100 Apis hives, with 50 hives near Austin, Texas, 25 near King Mongkut’s University of Technology Thonburi (KMUTT), and 25 near Chiang Mai University (CMU). The Thailand sites offer the advantage of year-round colony activity, while the Austin deployment provides essential U.S.-based validation and the potential for future collaboration with Texas A&M University entomologists. MeliponaShield™ will be tested exclusively in Melipona and Trigona hives near KMUTT, where the tropical climate and Dr. Orawan Duangphakdee’s internationally recognized stingless bee research program make it an ideal location. TransitShield™ units will be deployed during the transport of beehives to pollination sites.

Appendix A: Device Descriptions

Apiculture Technology International – CSM/CSMA-Based Systems

1. HiveShield™

Function: HiveShield™ is an internally installed environmental sensing system for Apis mellifera and Apis cerana colonies. It continuously monitors volatile organic compounds (VOCs), volatile sulfur compounds (VSCs), sound pressure levels (SPL), temperature, humidity, and atmospheric pressure to detect early signs of colony stress, brood decay, or environmental contamination.

Key Features:

Suspended between Langstroth frames or embedded into top bar

MOX-based gas sensors with temperature-cycled desorption and detection

Dual operation modes:

Baseline Trend Mode for long-term VOC drift detection

Inference Mode using machine-learning classification of compound fingerprints

BLE setup for initialization, GNSS tagging, and OTA firmware updates

LoRa telemetry for remote data transmission

Night-mode operation scheduling to reduce noise and preserve energy

Internal battery (≥6 month life), with optional solar top-up module

Purpose in Field Test: To evaluate the system’s ability to detect microbial or environmental stressors in managed Apis colonies and validate AI classification against biological inspections and laboratory results.

2. MeliponaShield™

Function: MeliponaShield™ is an external chemical and environmental monitoring system designed for stingless bee hives, including Melipona and Trigona species. It provides the same sensing capabilities as HiveShield™ but is adapted for compact hives where internal frame mounting is not feasible.

Key Features:

External mounting with passive air diffusion through protected intake ports

Supports both top-entry and side-mount configurations

Fully sealed, weather-resistant enclosure with hydrophobic gas-permeable membranes

Uses the same CSM/CSMA architecture as HiveShield™

Configurable sampling and reporting cycles

BLE + LoRa communication interface

Optimized for humid, forested environments typical of stingless bee habitats

Purpose in Field Test: To validate performance under tropical conditions and assess early detection of colony stress in native stingless bees without disrupting hive structure or behavior.

3. TransitShield™

Function: TransitShield™ is a field-deployed chemical sensing system designed to detect stressors during the transportation of hives. These stressors include out of range temperature and humidity, excess vibration and shock that cause hidden damage to the beehive.

Key Features:

Shared sensing platform (CSM/CSMA) with AI-driven classification

Continuous monitoring during transit interval

GNSS-tagged alert telemetry for traceability and spatial mapping

Configurable LoRa broadcast range and optional BLE setup for mobile app control

Purpose in Field Test: To confirm TrabsitShield’s ability to detect vibration and shock levels as well as temperature and humidity level events and provide actionable, location-tagged data for farmers, regulators, or researchers.

Appendix B – Program Costs

Appendix C – CSM/CSMA

Chemical Sensing Module and Apparatus for Beehive and Agricultural Monitoring

Technical Description and Deployment Framework

1. Introduction

The Chemical Sensing Module (CSM) and Chemical Sensing Module Apparatus (CSMA) form a unified sensing platform designed for continuous monitoring of volatile compounds in biologically sensitive environments such as beehives and agricultural zones. The system integrates multi-phase gas sensing protocols, embedded classification algorithms, and wireless communication interfaces into a compact, low-power unit suitable for long-term field deployment.

This platform supports both passive and active air sampling, compound-specific classification via temperature-cycled metal oxide (MOX) sensors, and context-specific operation modes tailored for internal hive environments, stingless bee colonies, and open-air agricultural settings.

2. System Overview

2.1. Chemical Sensing Module (CSM)

The CSM is a miniaturized, sensor-integrated subsystem comprising:

- One or more metal oxide semiconductor (MOX) gas sensors,

- A programmable heater driver,

- Signal acquisition circuitry (e.g., ADC, temperature/humidity compensators),

- A microcontroller for processing and control,

- Firmware for dynamic response profiling and AI inference,

- Interfaces for BLE and/or LoRa communication.

2.2. Chemical Sensing Module Apparatus (CSMA)

The CSMA is the physical embodiment of the CSM and includes:

- A protective mechanical housing (e.g., waterproof, thermally managed),

- Mounting interface (e.g., frame clip, hive entry port, external bracket),

- Embedded battery and optional solar charging,

- Communication subsystem (LoRa, BLE, or hybrid),

- Non-volatile storage for calibration data and historical logs.

The CSM is mounted within the CSMA using vibration-dampened standoffs and is thermally isolated from the external housing to maintain sensor accuracy.

3. Operational Modes

3.1. Baseline Trend Mode

In this mode, the sensor operates with a fixed heater temperature profile and captures long-term changes in gas-reactive resistance. It is ideal for:

- Detecting microbial fermentation (e.g., brood decay),

- Monitoring environmental deviation due to hive aging or stress,

- Supporting colony-level health profiling.

Algorithmic steps include:

1. Desorption phase (high-temp cleaning, ~320 °C)

2. Steady-state monitoring at optimized temperature (e.g., ~240–260 °C)

3. Compensation for ambient temperature and humidity

4. Comparison to normalized baseline using ΔR/R₀ detection

Baseline is dynamically adjusted based on nightly data to reduce false positives caused by short-term fluctuations.

3.2. Inference Mode

Inference Mode employs temperature-cycled sensing to produce a resistance "fingerprint" that captures the dynamic redox interaction of multiple gas species.

Sequence:

- High-temp desorption phase (purge)

- Stepwise temperature ramp: e.g., 160°C → 220°C → 280°C → 320°C

- Resistance sampled at each step with defined dwell/delay parameters

- Feature vector extraction (e.g., gradient, curve area, inflection points)

- Classification via embedded neural network, decision tree, or hybrid classifier

This enables compound-class-level classification, such as:

- Pesticide residue (e.g., neonicotinoids),

- Pathogen biomarkers (e.g., putrescine, hydrogen sulfide),

- Fermentation or environmental decay signals.

4. Calibration and Initialization Process

4.1. Factory Burn-In and Calibration

- 24–72 hour stabilization under inert or filtered ambient air.

- Exposure to known benign gas mixture (e.g., 100% nitrogen).

- Storage of resistance curves as factory calibration profile in non-volatile memory.

4.2. Field Installation and Adaptation

Upon deployment:

- Hive-specific or field-specific adaptation begins.

- Passive data collection over ~7 days under real conditions.

- Internal baseline profile generated for in-situ reference.

Dual-baseline methodology improves detection specificity by comparing:

- Long-term baseline drift (environmental/infrastructure),

- Real-time deviations (microbial onset, chemical exposure).

5. Telemetry and Configuration

5.1. Communication Protocols

- BLE for on-site configuration, firmware updates, and initial pairing.

- LoRa for remote data transmission in rural or distributed deployments.

- Optional dual-mode fallback (BLE > LoRa, or vice versa).

5.2. GNSS Tagging and Metadata Storage

- Device stores unique Hive ID and geolocation.

- Metadata included in all transmissions (timestamp, signal strength, error flags).

6. Power Management

- Primary power: Li-ion or LiFePO₄ battery pack (field replaceable).

- Low-power sleep mode between measurements (<50 µA standby).

- Measurement window: <60 seconds active per cycle.

- Optional: solar recharging for seasonal installations.

Measurement frequency and mode can be tuned based on:

- Battery level,

- Expected threat level (e.g., disease season),

- Hive species and behavior profile.

7. Environmental Resilience and Mounting

7.1. Mounting Configurations

- Apis hives: suspended between Langstroth frames.

- Stingless bee hives (Melipona, Trigona): top-entry or external mounts.

- Field deployments: post-mounted weather-sealed enclosures.

7.2. Protective Measures

- Hydrophobic, gas-permeable membrane over diffusion ports.

- Conformal coating of PCB (optional).

- Pressure equalization membrane (for rapid barometric changes).

8. AI and Machine Learning Compatibility

- Compatible with pre-trained classifiers embedded as lookup tables or binary models.

- Supported models: decision trees, SVMs, neural networks (quantized).

- Firmware allows OTA updates to replace or augment models.

- Feature extraction pipeline aligned with power-of-two node counts (16, 32, 64) for compressed inference.

- Models trained on both positive (known chemical classes) and negative (benign) data sets.

9. Applications

The CSM/CSMA platform supports deployment in:

- Honeybee hives for disease and stress detection (HiveShield™),

- Stingless bee colonies with compact or non-invasive needs (MeliponaShield™),

- Honeybee hives in transit for disease and stress detection (TransitShield™),

- Future modular platforms for soil gas sensing or VOC-based pathogen alerts.

10. Summary

The Chemical Sensing Module and its integrated Apparatus provide a unified platform for low-power, high-specificity environmental gas monitoring in biological and agricultural systems. It leverages modern sensor technology, AI classification, and scalable deployment architecture to meet the demands of beekeepers, researchers, and ecological stewards.

Its utility across multiple bee species (Apis mellifera, Apis cerana, Melipona, and Trigona), as well as its application in organic agriculture, makes it a cornerstone in next-generation epizootiology and environmental diagnostics.