The Download: monkey names, and smart masks for health monitoring

August 31, 2024

Two stories from MIT Technology Review's 'The Download' highlight opportunities for Nigeria's tech landscape. One shows how researchers used machine learning to identify specific vocal labels marmoset monkeys use to address each other, proving non-human primates use 'names'. The other describes a smart mask analysing breath biomarkers to monitor lung health non-invasively. Though developed in US labs, their principles—bio-inspired AI, continuous health monitoring, and interpretable machine learning—align with Nigerian needs. From agritech decoding animal behaviour to wearables managing urban respiratory conditions, these innovations offer transferable lessons for Nigerian shoppers, entrepreneurs, and policymakers at the biology-data-science intersection.

Key takeaways

The discovery that marmosets use specific vocalisations to 'name' companions offers a model for developing more transparent and interpretable AI systems, crucial for building trust in Nigerian financial, agricultural, and healthcare applications.
Smart masks capable of analysing breath biomarkers represent a significant step toward accessible, continuous preventative health monitoring—particularly relevant for managing respiratory conditions prevalent in Nigeria's urban and rural populations.
Both innovations underscore the power of bio-inspired design: looking to natural systems for solutions to technological challenges in AI interpretability and health diagnostics.
For Nigerian businesses and policymakers, the lessons emphasise localisation—adapting core technologies to local biological contexts, environmental conditions, affordability constraints, and user preferences to ensure relevance and impact.
Investing in explainable AI and preventative health tech now positions Nigerian stakeholders to meet rising global demands for transparency, accountability, and proactive wellness management.

What Marmoset 'Names' Reveal About the Future of AI Interpretability

The discovery that marmosets use distinct vocalisations to label companions challenges assumptions about human language uniqueness. By recording high-pitched chirps and applying pattern-recognition software, scientists found consistent acoustic signatures tied to specific monkeys. Playing these 'labels' back elicited stronger responses, indicating intentional, referential communication. This informs efforts to make AI systems more transparent. Current AI models, especially large language models in chatbots, often function as 'black boxes' with opaque decision-making. A parallel effort tweaks how artificial neurons combine signals in neural networks to create AI whose logic is easier to trace and verify. For Nigerian businesses using AI in credit scoring, agricultural forecasting, or customer service, interpretability builds trust, meets regulatory expectations, and identifies biases. Imagine a Kano microfinance AI explaining loan declines with verifiable factors, or an Oyo State crop disease system showing farmers why it flagged leaf discolouration as pathogenic. The marmoset study shows complex communication—biological or artificial—can be decoded with the right tools. Investing in explainable AI (XAI) now positions Nigerian tech firms ahead of global curves where accountability and transparency are competitive differentiators. Furthermore, understanding animal communication has agritech applications: monitoring livestock vocalisations for distress or illness. The marmoset method—long-term audio recording with machine learning pattern recognition—could monitor poultry health or detect cattle stress, reducing losses and improving welfare.

Smart Masks and the Rise of Continuous, Non-Invasive Health Monitoring

The smart mask analysing exhaled breath for biomarkers addresses a preventive healthcare gap: continuous, convenient physiological monitoring outside clinics. Exhaled breath contains volatile organic compounds (VOCs) and molecules indicating inflammation, infection, or metabolic disorders. Wearable sensors detecting these substances enable real-time lung health assessment. For Nigeria, where respiratory illnesses contribute significantly to morbidity and mortality—exacerbated by Lagos and Port Harcourt air pollution, harmattan dust, and smoking—this technology holds promise. Imagine an asthma patient using a low-cost mask to track exposure triggers and lung function trends, sharing anonymised data with clinicians for proactive treatment adjustment. Or rural community health workers using aggregated mask data to identify emerging respiratory outbreaks before clinics are overwhelmed. Business model potential is compelling: shifting from periodic hospital visits to prevention and early intervention. Nigerian health-tech startups could develop localized masks for tuberculosis, COPD, or asthma exacerbations, partnering with telemedicine platforms for timely consultations based on alert thresholds. Emphasis on non-invasiveness and home use aligns with preferences for convenient, stigma-free health management. Aggregated, anonymised data from widespread adoption could offer public health insights: Abuja city planners correlating mask-derived pollution exposure with traffic patterns to inform urban design, or national health agencies monitoring regional biomarker variations to allocate resources during seasonal health challenges. Success requires affordable, durable designs (for dust, humidity) and clear data privacy frameworks to build trust.

Broader Implications: Bio-Inspiration, Localisation, and Nigeria's Tech Strategy

These stories converge on bio-inspiration's value: decoding primate vocalisations to improve AI or harnessing breath's diagnostic potential. Innovation emerges at biology, data science, and engineering's intersection. For Nigeria, this reinforces that effective technology adoption begins with understanding local biological, environmental, and social contexts—not just importing finished products. Consider AI bias: models trained on temperate climate or specific ethnic data may misdiagnose skin conditions, misunderstand speech accents, or misinterpret agricultural imagery in Nigerian settings. The marmoset research reminds us meaningful patterns exist in all communication; the task is to listen carefully and build tools attuned to local 'languages'—vocal, visual, or behavioural. Similarly, health monitors must account for prevalent local pathogens, environmental exposures, and genetic factors. Localisation extends to business models and user experience. A Silicon Valley-priced smart mask will have limited uptake in Nigeria. Success requires re-engineering for affordability via locally sourced components or simpler sensor arrays focused on prevalent health concerns. Distribution could leverage patent medicine vendors or community health worker networks instead of relying solely on e-commerce. User interfaces should prioritise clarity and actionability over technical sophistication, using voice prompts in local languages or simple colour-coded alerts. These innovations highlight opportunities in Nigerian tech education and entrepreneurship. Universities and polytechnics could develop courses in bio-inspired engineering, sensor development, or explainable AI. Incubators and hubs should seek founders working on agritech decoding animal behaviour or health wearables for tropical environments. Government research grants and innovation funds could prioritise projects demonstrating clear local impact pathways, leveraging Nigeria's biodiversity and demographic profile as assets.

What this means

The stories of naming monkeys and smart masks monitoring breath may seem distant from Nigerian market realities, but they encapsulate fundamental shifts in how technology interacts with biology and human behaviour. As AI systems become more pervasive in cross-sector decision-making, demand for understandable, trustworthy tools will grow. Simultaneously, the burden of preventable disease underscores the value of technologies empowering individuals to monitor health simply and continuously. For Nigerian entrepreneurs, the message is clear: innovation does not always require inventing something entirely new. Often, it involves observing nature's solutions—a monkey's call or breath chemistry—and adapting their principles to solve local problems with ingenuity and respect for context. By embracing bio-inspired thinking, prioritising explainability and accessibility, and grounding innovation in local realities, Nigeria can participate in and shape the global tech evolution to reflect its unique strengths and needs.

Originally reported by technologyreview.com. Adapted for our readers with AI assistance.