Page 69 - Read Online

Ganesan et al. Art Int Surg 2024;4:364-75 https://dx.doi.org/10.20517/ais.2024.68 Page 372

As AI aims to improve prognostic prediction, providers should be aware of the challenges associated with
communicating poor prognosis to patients. Providers might become fatigued when addressing treatment
options for patients with unfavorable prognoses. Patients may question the validity of AI, and providers
should be prepared to have such discussions. Although AI is prepared to analyze infinite amounts of data
and suggest prognoses, providers must be equipped to discuss all those findings comprehensively.

Current gaps and the future of AI in wound care
Thus far, AI has been leveraged to process large amounts of data quickly and accurately, proving useful for
wound diagnosis and characterization, management, treatment, and prognosis prediction. With
advancements in AI-assisted technology, considerations for equitable access must be addressed.
Understanding how this technology will be funded, whether through insurance, individual payers,
government and public funding, or hybrid models, is crucial for equitable access. Wound management
itself, even without AI-assisted technology, is expensive, and there are a plethora of available options for
dressings, antibiotics, and more. Powerful and detailed AI algorithms could be used to help sort which
methods of management might be most cost-effective given a patient’s insurer. Utilizing the most cost-
effective methods of management from the onset of wound diagnosis could help save on downstream costs.

Additionally, integrating developed smartphone apps into clinical practice, rather than just a trial setting,
should be studied. Considering the incorporation of AI into digital platforms used by healthcare providers,
such as EMRs, may allow for real-time wound analysis. If AI technology can be implemented in rural areas,
providers might be able to guide remote wound care. However, once AI is integrated into these settings,
identifying who will supervise the data, whether the provider, hospital, or a third-party data analytics group,
will be paramount to seamlessly incorporating accurate and accessible AI-assisted technology into wound
care.

Of the current studies on AI advancements in wound care, very few report demographic data. Fewer
reported AI-assisted technology’s accuracy, sensitivity, and specificity stratified by racial background. This
is particularly important for image-based detection methods, where an accurate AI-assisted technology
should be able to adequately diagnose wounds regardless of skin color. Further reporting on demographics
will improve transparency and reduce bias from AI-assisted technology. Diversifying datasets for AI
training will also ensure less biased data and improve output accuracy.

Of note, patients from diverse backgrounds may heal in clinically different ways. Keloids are more likely to
[52]
develop after injury in those of African and Asian ancestry . Hypertrophic scarring is more likely to occur
[53]
in those with darker skin colors . Not only will demographically diverse data provide insights into these
conditions, but AI may be able to predict when these complications might occur. Kim et al. used a neural
network structure along with multinominal logistic regression to identify that scar severity was positively
associated with postoperative itching and pain . They found that postoperative adhesion/tightening and
[54]
induration/edema were negatively correlated with scar severity in patients. More research must be done to
further predict keloid and hypertrophic scarring development in patients.

Additional research can be done into predicting wound healing complications such as sepsis and
necrotizing fasciitis. Although AI’s prognostic ability for both cases has been studied, they have not been
studied in the context of wound healing [55-57] . AI can be leveraged in molecular biology as well. So far, most
of the technologies that analyze wounds have focused on the wound itself. However, wounds are often
accompanied by a heterogeneous array of exudates, calluses, edema, maceration, and excoriations. Wound
healing is often impacted by the specific type and amount of bacteria that are in the wound. Research into