Permission to Advance: The Role of AI and Digital Health Tools in the Identification of the Trafficked

In 2023, the National Human Trafficking Hotline identified 9,619 human trafficking cases in the United States. Further investigation revealed that these cases involved a total of 16,999 victims. The United States consistently ranks high in the number of human trafficking cases reported annually, the crime remaining a profitable and pervasive industry. According to the International Labour Organization, an estimated 27.6 million people were trapped in “modern slavery” in 2021 – a number defined by the regulations placed under US law. Unfortunately, this atrocious number has only increased since 2021 and remains prevalently high in the United States as a very profitable business.

But what exactly is human trafficking? Human trafficking is a crime that compels individuals to labor or to engage in a commercial sex act, rooted in exploitation of the individual and the goods they provide, rather than a country or border (U.S. Department of State, 2024). The signs of human trafficking are not always easy to identify. When people think of human trafficking, they often think about a situation where someone is kidnapped and forced into labor. They may think of a situation like Samirah and Enung’s, who were hired out of Indonesia to work at a rate of $100-$200 per month (Logan et al., 2009). But once they reached their employer, they were stripped of their forms of identification and were starved, forced to work, abused, and kept locked in the house. While the stereotypical image of human trafficking often involves cases that incorporate abduction and physical captivity such as Samirah and Enung’s and are deeply important to understand and identify, it is crucial to understand that human trafficking does not always involve physical kidnapping. Many victims are coerced or manipulated in ways that are not always easy to identify or see but are still a clear violation of their freedom.

For instance, grooming is a form of human trafficking that involves the psychological manipulation of an individual by creating trust between the perpetrator and the victim and the perpetrators can be as close as a family member (Polaris, 2021). It’s hard for outsiders to isolate this behavior as trafficking or grooming, especially because it is occurring within family dynamics. A significant proportion of the estimated 27.6 million trafficked individuals are not physically abducted but are instead manipulated and exploited while living seemingly ordinary lives.

As a result, healthcare providers have a crucial role to play in the identification of the trafficked (Sutherland, 2019). Research indicates that many trafficked individuals access healthcare during exploitation, presenting an opportunity for intervention. Recognizing patterns in behavior and biometrics could be key to identifying victims, bringing into conversation the use of artificial intelligence to aid healthcare providers in recognizing the trafficked.

Artificial intelligence (AI) is a type of computer science that learns from experiences and adapts to new inputs to better predict outcomes (UIUC, 2024). Some types of artificial intelligence include machine learning (which means that the algorithm learns from the experience), pattern recognition, and decision making. For this particular situation, pattern recognition is the main focus and its potential utilization in healthcare settings. Between 2018 to 2022, a software application called “Octavia” was implemented in three different hospitals under the CommonSpiritWealth nonprofit organization (Duke et al., 2023). Octavia analyzed patients’ electronic health records (EHRs)—containing details such as demographics, treatment history, and medical data—four times per hour. It then categorized patients into clinically relevant phenotypic groups. After this initial analysis, Octavia was able to identify potential cases of human trafficking at a rate of 1-8 daily, then of which a HRPN (High- Risk Patient Navigator) screened about 44% of them and confirmed the status of the patient as a trafficked victim at 23% of those screened. This number represents cases that would've otherwise been missed by the healthcare sector and suggests the ever-evolving use of AI in this field of trafficking identification.

Beyond biometrics, AI systems can also analyze keywords and behavioral cues to identify potential victims. Pattern recognition software has the capability to automatically recognize patterns in data and has the capacity to learn from past mistakes to increase accuracy in classification (Serey et al., 2023). There has been advancing research in pattern recognition in relation to human-level concept learning. For example, CLEVR (Compositional Language and Elementary Visual Reasoning) provides insights into complex visual and linguistic patterns to classify visual datasets into categories (Holzinger, 2023). CLEVRER-Humans takes into account phrases (and other auditory noises) such as “because” and “responsible for" to define situations where someone is describing a chain of events or a sequence. Using a similar frame but for the trafficked, certain combinations of words can be used by software similar to CLEVRER-Humans to help healthcare providers determine the presence of the trafficked as they are treating them. This could be a huge advance in the field of liberating trafficked and in technology.

While the potential of AI is immense, its application in healthcare and trafficking detection raises ethical concerns. Safeguarding patient privacy and ensuring unbiased algorithmic design are of utmost concern. Additionally, the effectiveness of AI depends on robust data and comprehensive training. Trafficking victims who exhibit atypical patterns or lack clear documentation could still be overlooked.

Integrating AI into healthcare for trafficking detection could be life-changing, offering a powerful tool to recognize victims and intervene in their exploitation. By refining technologies like Octavia and CLEVRER-Humans, healthcare systems can increase their capacity to detect and assist trafficked individuals. However, collaboration between healthcare providers, policymakers, and technology developers is important to ensure ethical and effective implementation.

Ultimately, using healthcare and AI to combat human trafficking represents a trailblazing shift in the fight against this global issue. As technology evolves, so too must our commitment to protecting the vulnerable.