The invasive Anopheles stephensi species is of profound global health concern due to its efficient transmission of malaria, particularly in urbanized landscapes. Unlike native African Anopheles mosquitoes that predominantly breed in natural water bodies, An. stephensi thrives in artificial containers, such as discarded tires and buckets, creating unique challenges for disease containment. This adaptation has enabled the species to expand its reach across rapidly urbanizing regions, putting an additional estimated 126 million people in Africa at heightened risk of malaria infection.

The identification of this elusive vector in Madagascar was made possible by a single citizen-scientist-submitted image via NASA’s GLOBE Observer app, a platform empowering the public to contribute valuable scientific data through smartphone technology. The image—depicting a mosquito larva discovered in a tire—was subjected to AI-based image recognition algorithms, meticulously trained on thousands of authenticated mosquito images. Impressively, the algorithm classified the larva as Anopheles stephensi with an accuracy exceeding 99%, a testament to the maturation of machine learning in ecological and epidemiological applications.

This breakthrough detection signifies more than a singular discovery; it exemplifies a paradigm shift in public health surveillance. Traditional entomological monitoring methods often involve labor-intensive field collections and delayed laboratory analysis, leaving critical gaps where invasive species can establish and expand unnoticed. The integration of AI-driven diagnostics coupled with widespread public engagement through mobile applications dramatically accelerates detection timelines, enabling near-real-time monitoring of disease vectors on a scale previously unattainable.

Despite the inability to perform genetic confirmation—owing to the immediate destruction of the specimens post-collection—the consistency of findings, including observations of over 100 other Anopheles larvae in similar breeding sites on the same day, lends considerable weight to the identification. Notably, the same year as the discovery coincided with a precipitous doubling in malaria cases and fatalities in Madagascar, paralleling global concerns about the rapid spread of An. stephensi in new ecological niches.

This study is not solely about geographic discovery; it portends a looming public health crisis with direct implications beyond the African continent. In the United States, where malaria was long considered eradicated in local transmission contexts, 2023 witnessed localized outbreaks for the first time in over two decades. Florida emerged as a hotspot, reporting more cases than all other states combined, starkly highlighting the urgency of innovative surveillance tools.

Central to this initiative is the development of next-generation AI models that replicate the functionality of facial recognition technologies, but for mosquito larvae and adults. By harnessing extensive datasets of labeled mosquito imagery, these models not only distinguish between species with high precision but also facilitate scalable, cost-effective surveillance by non-experts. Such technological sophistication offers a complementary approach to genomic techniques and traditional surveillance, especially in resource-limited settings.

The multi-disciplinary collaboration at USF, spanning departments of Integrative Biology, Artificial Intelligence, Cybersecurity and Computing, and Public Health, reflects the complex intersection of ecology, technology, and health sciences necessary to confront mosquito-borne diseases. Supported by grants from the National Institutes of Health and National Science Foundation, this work builds upon prior award-winning research by the team and sets the stage for advanced hardware innovation, such as AI-enabled smart traps designed to automatically identify multiple mosquito species in situ.

The strategic vision articulated by the researchers envisions a future where smart traps equipped with embedded AI analyses serve as sentinel devices across urban and rural landscapes. These traps could autonomously transmit data on mosquito presence and species composition, thereby informing timely public health interventions, vector control strategies, and epidemiological modeling. This represents a proactive leap in disease control infrastructures, combining engineering and bioinformatics to outpace vector adaptation and spread.

Reflecting on the broader significance, Dr. Carney emphasized that although mosquitoes are commonly perceived as mere nuisances, a small fraction—approximately 3%—are vectors for human diseases. The deployment of citizen science applications paired with AI analytics empowers communities globally, enabling them to partake actively in identifying and mitigating threats posed by these disease vectors. This democratization of surveillance holds immense promise for enhancing public health resilience.

Dr. Chellappan added that the rising prominence of AI in public health domains cannot be overstated, particularly in mosquito surveillance where rapid identification and response are critical. The innovations spearheaded by the USF team are poised to transform epidemiological landscapes worldwide, fostering early-warning systems that can curb outbreaks before they proliferate extensively.

In sum, the USF-led initiative marks a seminal step in harnessing artificial intelligence and public participation to confront the complexities of invasive mosquito species and their attendant health risks. The convergence of AI, mobile technology, and community engagement charts a promising trajectory toward smarter, faster, and more inclusive global disease surveillance frameworks, offering hope in the ongoing battle against malaria and other mosquito-borne illnesses.

Subject of Research: Animals

Article Title: Artificial Intelligence and Citizen Science as a Tool for Global Mosquito Surveillance: Madagascar Case Study

News Publication Date: 28-Oct-2025

Web References:

• NASA’s GLOBE Observer app: https://observer.globe.gov/about/get-the-app
• Journal Insects: https://www.mdpi.com/journal/insects
• USF Faculty Ryan Carney: https://www.usf.edu/arts-sciences/departments/ib/people/faculty/ryan-carney.aspx
• USF Faculty Sriram Chellappan: https://www.usf.edu/ai-cybersecurity-computing/people/faculty/chellappan-sriram.aspx

References:

• Carney, R., Chellappan, S., et al. (2025). Artificial Intelligence and Citizen Science as a Tool for Global Mosquito Surveillance: Madagascar Case Study. Insects.
• Prior study: https://www.mdpi.com/2075-4450/13/8/675

Image Credits: Ryan Carney, University of South Florida (Credit: USF)

Keywords: Malaria, Parasitic diseases, Infectious diseases, Diseases and disorders, Public health

Diagnosing PTSD in children often relies heavily on self-reported questionnaires and clinical interviews, both of which can be significantly limited by a child’s cognitive development, emotional awareness, and communication abilities. Young children may not have the language or the emotional framework to adequately articulate their feelings and experiences, which can lead to underdiagnosis or misdiagnosis. This is where the USF team’s efforts draw attention. By harnessing AI and machine learning, they aim to overcome these limitations and provide clinicians with a richer, more reliable understanding of a child’s mental health.

At the foundation of this research lies an innovative concept introduced by Salloum, who observed how children’s facial expressions changed dramatically during trauma interviews. These instances showed more than words ever could, revealing the depths of the emotional turmoil they were experiencing. Recognizing the potential of AI to capture these nuanced expressions, Salloum approached Canavan to explore the possibility of using technology to quantify these observable cues in a structured manner that respects the privacy of the children involved.

Canavan’s expertise in facial analysis and emotion recognition led him to repurpose existing technological tools from his lab to create a system that prioritizes patient privacy at every stage. The AI developed by the team does not utilize raw video footage—rather, it anonymizes and de-identifies data. By focusing solely on facial movements, such as eye gaze and head position, while filtering out identifying information, the researchers can analyze vital emotional cues without compromising each child’s privacy.

In developing their methodology, the researchers built an extensive dataset derived from 18 sessions featuring children recounting their emotional experiences. This dataset contains more than 100 minutes of video footage for each child, categorizing roughly 185,000 individual frames packed with data on subtle facial movements linked to varied emotional expressions. The results were promising; the AI successfully detected distinct patterns in the facial movements of children diagnosed with PTSD, providing insight into how their expressions during therapy sessions differed from those during conversations with their parents.

The researchers noted that clinician-led interviews elicited more revealing emotional responses than interactions with parents. This finding is particularly significant, as it correlates with existing psychological literature which posits that children may feel more comfortable being emotionally expressive in the presence of therapeutic professionals. These insights suggest that the AI system can serve not merely as a diagnostic tool, but as an invaluable adjunct to therapist interventions, potentially enhancing therapeutic outcomes by offering real-time feedback.

As this research progresses, the team is keen to examine various factors that might influence their findings, including the roles of gender, culture, and age in facial expression analysis. Special emphasis will be placed on preschoolers, as they represent a challenging demographic where verbal communication is often limited and diagnoses depend largely on parental observations. By expanding their research scope, the team seeks to ensure that their AI tool is both comprehensive and free from biases, further solidifying the ethical standards of their work.

Though still in its nascent stages, the implications of this research could be profound, offering a transformative shift in the landscape of child mental health diagnosis. Many participants in their ongoing studies have shown complex clinical profiles, including co-occurring conditions such as depression and anxiety, attesting to the real-world applicability and potential accuracy of the AI system. Conducting a study with such ethically sound practices is indeed a rare achievement, especially when the subjects involved are vulnerable populations, making this research paradigm particularly noteworthy.

The implications extend beyond mere diagnostics. If this new methodology showcases efficacy in larger clinical trials, it could redefine conventional approaches to diagnosing and treating PTSD in children. By utilizing common tools such as video and artificial intelligence, mental health care could evolve into a future where diagnostics are more precise, less traumatic, and ultimately, more effective in rendering help when it is needed most.

In advocating for a future where mental health care is significantly improved, researchers like Salloum and Canavan are paving the way for a nuanced understanding of emotional expression in children. By integrating innovative technology and clinical acumen, they provide hope for better recognition and treatment pathways for children suffering from the debilitating effects of trauma, shaping a landscape where mental health diagnoses are informed not just by words, but by the intricate language of nonverbal cues.

As the field of child mental health continues to evolve, the commitment of researchers at the University of South Florida marks a pivotal moment in understanding, diagnosing, and treating PTSD. With the potential to change how therapists engage with young patients, their work champions a future where technology serves as an ally to the human touch needed in therapeutic settings.

Through this transformative lens, it is clear that the intersection of artificial intelligence and childhood trauma research holds tantalizing possibilities not only for improving diagnostics but for reshaping the entire landscape of mental health care for children. As further research unfolds, there is anticipation that this innovative methodology may one day lead to broader acceptance and integration of AI technologies in clinical practice, enhancing therapists’ capacity to support and heal some of society’s most vulnerable members.

Subject of Research: Children with Post-Traumatic Stress Disorder
Article Title: Multimodal, context-based dataset of children with Post Traumatic Stress Disorder
News Publication Date: 30-June-2025
Web References: https://www.usf.edu/index.aspx
References: To be determined upon peer reviews and publication
Image Credits: Credit: USF

Keywords

The study titled “The Illusion of Authenticity in Online Reviews: Truth Bias and the Role of Valence,” recently published in the leading journal Information Systems Research, addresses a pressing question within consumer behavior literature and digital marketing: Are individuals inherently skeptical toward the authenticity of online reviews, or do they exhibit a cognitive bias that compels them to trust information unless challenged by undeniable proof? The investigation uncovers the pervasive influence of what psychologists call the “truth bias,” the innate predisposition to believe communicated information is truthful unless there is strong indication suggesting otherwise.

Associate Professor Dezhi Yin of the University of South Florida’s Muma College of Business, a principal investigator in the project, explains, “Our objective was to deepen the understanding of how consumers discern between genuine and fabricated online reviews. While prior studies have typically focused on fake review detection via technical algorithms, we wanted to center on the consumer perspective—who remains the ultimate target and victim of review manipulations.” The research team also includes co-authors Samuel D. Bond and Han Zhang from Georgia Institute of Technology, with Han Zhang also affiliated with Hong Kong Baptist University, combining expertise in information systems, behavioral science, and marketing.

Over a five-year period from 2018 to 2023, the researchers conducted a series of five experimental studies involving hundreds of participants. These participants reviewed collections of online consumer reviews—predominantly for restaurants—and were asked to classify each review as either “real” or “fake.” Crucially, participants were informed beforehand that half of the reviews they were about to assess were fabricated, establishing an environment of guarded skepticism. Contrary to expectations, however, participants still overwhelmingly categorized a majority of the reviews as authentic, revealing a pronounced truth bias that blurred their ability to detect deceptions effectively.

One revealing part of the study involved participants viewing twenty restaurant reviews simultaneously, with explicit knowledge that only half represented genuine consumer opinions. The design allowed participants to switch between reviews freely, providing the opportunity to recalibrate their judgments by comparing across the entire dataset. Yet, the average number of reviews classified as authentic exceeded the actual number of genuine reviews, indicating persistent over-trust. This tendency underscores the cognitive difficulty humans face when counteracting inherent biases even when they are primed to be skeptical.

An intriguing facet of the research explored the impact of “valence,” or the positive or negative tone of reviews, on perceptions of authenticity. Real-world analyses indicate that fake reviews tend to skew negative more significantly than positive, as competitors often seek to undermine rivals by posting unfavorable content. Paradoxically, study participants demonstrated a stronger propensity to believe negative reviews over positive ones. This disconnect between actual deceptive behavior and consumer perception highlights a critical vulnerability in the way online reputations are shaped and maintained.

The consequences of this truth bias extend beyond academic interest to practical challenges for online platforms, e-commerce marketplaces, and businesses reliant on consumer feedback. Since users inherently trust most reviews—even when warned about fabricated content—platforms that depend on users reporting fake reviews face an uphill battle. Temporal delays in flagging false content and users’ cognitive biases reduce the overall effectiveness of community-driven policing. Therefore, the authors advocate for proactive and systematic detection mechanisms using advanced algorithms and machine learning models targeted especially at negative reviews, given their outsized influence and susceptibility to fraud.

Furthermore, the study recommends platforms consider thoughtful user interface modifications to help consumers navigate the complex landscape of online reviews more accurately. Design choices such as segregating positive and negative reviews, integrating rating-based sorting, and highlighting suspicious patterns could enhance users’ decision-making faculties and reduce deception’s impact. These strategies offer scalable and user-friendly approaches to complement technological detection, fostering a healthier ecosystem of consumer information exchange.

Beyond immediate practical applications, this research opens avenues for deeper interdisciplinary inquiry intersecting psychology, marketing, information systems, and digital communication. Understanding the interplay between cognitive biases such as truth bias and the contextual factors shaping consumer judgments can inform more robust models of online behavior and misinformation dynamics in the digital economy. Such insights are pivotal for devising effective interventions to combat misinformation and restore trust.

As the digital marketplace continues to evolve, the findings of this study serve as a timely reminder that the battle against misinformation is not solely technical but fundamentally human. Consumer psychology wields considerable influence in how information is received and acted upon, often overriding rational considerations. By addressing these psychological dimensions, researchers and practitioners can develop more nuanced strategies to protect consumers and maintain the integrity of online information.

In conclusion, while increased awareness about fake reviews does prompt some caution, this research illuminates the enduring and potent nature of the truth bias in consumers’ evaluations of online reviews. Platforms and marketers must recognize this psychological tendency to design more effective countermeasures, balancing technological innovation with cognitive insights. Such combined efforts are essential to safeguard the veracity of consumer feedback and support informed, confident decision-making in an era where trust increasingly shapes digital interactions.

Subject of Research: People

Article Title: The Illusion of Authenticity in Online Reviews: Truth Bias and the Role of Valence

News Publication Date: June 5, 2025

Web References: https://pubsonline.informs.org/doi/10.1287/isre.2023.0339

Image Credits: Credit: USF

Keywords: Marketing research, Marketing, Advertising

Hydrogen, while increasingly heralded as a clean and efficient energy carrier, presents formidable safety challenges due to its flammability and propensity to accumulate in confined spaces where leaks can easily go unnoticed. Traditional hydrogen detection methods often require close proximity to the source, increasing risk and limiting overall effectiveness. The newly developed Raman analyzer, however, circumvents these limitations by employing an advanced optical technique that identifies and quantifies hydrogen molecules without the need for direct contact, thereby enabling safer, more efficient monitoring over larger distances.

The underlying principle of this device is Raman scattering, an optical phenomenon where incident light interacts with molecular vibrations within a material, producing scattered light with unique spectral fingerprints indicative of specific substances. While Raman spectroscopy is well-established in analyzing liquids and solids, leveraging it to detect trace gas concentrations, especially outdoors, has historically been hampered by the inherently weak nature of scattered signals and susceptibility to environmental interference. Overcoming these obstacles required meticulous engineering optimizations and innovative enhancements that elevate the technique’s sensitivity and robustness.

Central to this breakthrough is the application of multipass cavity enhancement, a sophisticated approach designed to amplify the Raman signal by reflecting the interrogation laser beam multiple times through the air sample, effectively increasing interaction length without compromising instrument stability. Unlike conventional multipass designs sensitive to alignment drifts caused by temperature fluctuations and mechanical vibrations, this analyzer’s cavity geometry is engineered to maintain alignment autonomously, ensuring consistent performance across varying environmental conditions.

The instrument harnesses a high-power laser emitting at 442 nanometers with a narrowly defined spectral linewidth of less than 0.1 nanometers, emitting several watts of power necessary to generate sufficiently strong Raman signals from trace amounts of hydrogen gas. Coupled with a highly sensitive spectrometer consuming under 10 watts of power, the system balances high performance with portability and energy efficiency, making it suitable for field deployment in remote or challenging locations.

Graduate researcher Charuka Arachchige, who led extensive field testing, reports that the analyzer maintained exceptional sensitivity and stability across diverse settings, including controlled laboratory rooms, spacious atriums, and fully open outdoor environments. The testing encompassed spatial mapping of hydrogen concentration gradients extending from a controlled source situated multiple meters away, where the instrument successfully distinguished hydrogen at levels as low as 63 parts per billion above ambient background—a detection threshold previously unattainable with portable technologies.

This capability not only enables rapid identification of even small leaks but also paves the way for detailed environmental assessments of natural hydrogen emissions. Researchers postulate that due to hydrogen’s geological production in subterranean reservoirs, the analyzer might serve as a powerful tool for exploiting these untapped energy resources, offering both safety monitoring and resource exploration benefits.

The innovation holds promise beyond industrial and geological applications. Given its ability to detect trace analytes with high specificity and sensitivity, the Raman analyzer could be adapted for a variety of medical diagnostics, environmental sensing, and chemical detection tasks where precision and portability are paramount. Its non-contact detection methodology also minimizes contamination risks, an essential consideration for biomedical and environmental fields.

The team’s success is rooted in overcoming long-standing challenges that have limited previous portable Raman gas analyzers, primarily related to stability, power consumption, and environmental adaptability. The engineered multipass cavity uniquely balances high finesse with tolerance to mechanical and thermal disturbances, eliminating the need for complex stabilization hardware that typically impedes portability. Combined with an optimized spectral acquisition protocol, the system achieves rapid measurement cycles, enabling near-real-time monitoring.

Moreover, the researchers are actively refining key performance metrics, including lowering the limit of detection further, accelerating data acquisition speeds, and miniaturizing instrument components without sacrificing sensitivity. These improvements aim to facilitate broader adoption in field applications and integration into commercial safety solutions and environmental monitoring networks.

The research, detailed in the forthcoming Applied Optics journal, demonstrates a compelling synergy between fundamental optical physics, advanced engineering, and practical application targeting the evolving needs of hydrogen economy safety and beyond. By translating precise Raman spectroscopic analysis into a durable and portable instrument, the team addresses a critical gap in monitoring capability that is increasingly urgent as hydrogen infrastructure expands globally.

For industries reliant on hydrogen, including transportation, power generation, and chemical manufacturing, this development offers a vital new dimension of proactive risk management. Early and accurate detection of leaks not only protects human safety and property but also curtails energy losses and environmental impact. Such real-world benefits underscore the transformative potential of integrating cutting-edge optical technologies into routine operational practices.

In conclusion, the portable Raman analyzer reflects a leap forward in trace gas detection technology, combining scientific ingenuity with practical utility. Its ability to remotely sense hydrogen at unprecedented low concentrations in ambient air under variable environmental conditions positions it as an essential tool in securing the safe proliferation of hydrogen as a sustainable energy vector.

Subject of Research: Portable Raman spectroscopy for trace hydrogen gas detection

Article Title: Portable Raman Hydrogen Concentration Mapping with Parts-per-Billion Sensitivity

News Publication Date: 24-April-2025

Web References:
DOI: 10.1364/AO.558965
University of South Florida: https://www.usf.edu/
Optica Publishing Group: https://opg.optica.org/
Applied Optics journal: http://opg.optica.org/ao

References:
C.M Arachchige, A. Muller, “Portable Raman Hydrogen Concentration Mapping with Parts-per-Billion Sensitivity,” Applied Optics, vol. 64, 2025.

Image Credits: Andreas Muller, University of South Florida

Keywords

Hydrogen, Chemical analysis, Environmental methods, Photonics