It's Time to Define the Global Burden of Velopharyngeal Insufficiency

Abstract

Velopharyngeal insufficiency (VPI) predominantly affects children with cleft palate, undermining their ability to communicate. As a result, intelligible speech generation is one of the most important outcomes following cleft palate repair. In low- and middle-income countries (LMICs), the elevated incidence of cleft palate, unavailability of speech services, and suboptimal surgical outcomes has contributed to a substantial yet poorly defined global burden of VPI. Tracking speech outcomes in LMICs is essential to assessing VPI severity and identifying patients needing care. Artificial intelligence and machine learning are well-suited to accommodate this goal.

Effective communication is a cornerstone of human existence, enabling individuals to form meaningful relationships, engage in society, and achieve personal and professional success. For individuals with velopharyngeal insufficiency (VPI), these aspects of social life are profoundly disrupted.^1-4 Velopharyngeal insufficiency impairs a patient's ability to communicate, leading to stigmatization, ostracization, and physical and psychosocial sequelae.^1-3 The burden of VPI is most pronounced during childhood, a time during which youth already face heightened risks of bullying, diminished self-esteem, anxiety, and depression.^2-5 These psychosocial challenges are compounded by the fact that VPI predominantly affects children with cleft palate, which carries its own distinct functional and psychological burdens.^6,7 As a result, one of the most important outcomes of cleft palate repair is generating intelligible speech. In the United States and other high-income countries, the risk of VPI ranges from 15% to 30% after palatoplasty.^8-10 Although 80% of surgically treated patients improve their speech, with 70% achieving near-normal speech, VPI remains a difficult condition to diagnose and treat.^11,12 This editorial highlights the challenges of identifying VPI, particularly in low- and middle-income countries (LMICs), and considers how artificial intelligence (AI) can revolutionize diagnostic practices and improve access to care.

In LMICs, socioeconomic and geospatial barriers complicate cleft care delivery and often result in poor surgical outcomes.^13-16 The incidence of cleft palate is higher, cleft treatment centers are fragmented, if not completely absent, and services beyond primary cleft lip and palate repair are typically unavailable.^17,18 In LMICs, a critical, yet often unavailable part of cleft care is that of speech-language pathology.^17,18 These factors have contributed to a significant global burden of VPI, with its true prevalence remaining unknown. Even when speech services are available, the cost of diagnosing and treating VPI can consume a considerable portion of a family's income, with out-of-pocket expenses for consultations, procedures, and follow-up care being disproportionately high, relative to the country's gross national income per capita.^19-21 This financial strain commonly forces families to forego necessary treatment, perpetuating cycles of underdiagnosis and inadequate care. Geographic barriers further worsen these inequities, as long travel distances and the scarcity of specialized providers in rural or remote areas can prohibit access to care.^22,23

The diagnosis of VPI in LMICs is further hindered by a convergence of systemic and individual barriers. Accurate diagnosis relies on a multidisciplinary team, including speech-language pathologists (SLPs), otolaryngologists, and plastic surgeons. Advanced adjunct diagnostic tools such as videonasoendoscopy play an important role in diagnosis and procedural selection, as well.^2,3,5,13,14 However, in many LMICs, chronic underinvestment in healthcare infrastructure severely limits the availability of these trained specialists and essential diagnostic equipment.^24,25 The scarcity of training programs for healthcare professionals in these regions exacerbates these problems, as local providers commonly lack the expertise to diagnose and manage VPI. Furthermore, many LMICs face a significant “brain drain,” with trained specialists commonly emigrating to higher-income countries in search of better opportunities, leaving underserved populations without access to critical expertise. ²⁶

Addressing these complex challenges requires innovative thinking. At present, no sustainable mechanisms exist to deliver VPI care for orofacial cleft patients in LMICs, and no estimates of global VPI prevalence are available. The first and most critical step is establishing screening mechanisms to accurately assess the prevalence and impact of VPI across different populations globally. This involves the implementation of comprehensive and standardized protocols for identifying new and postoperative cases of VPI. Furthermore, targeted screening efforts should focus on regions with a high incidence of orofacial clefts and significant barriers to healthcare access, ensuring inclusivity. Many outreach medical programs have instituted robust data tracking systems to ensure patient follow-up in even the most austere settings. Employed techniques range from paper chart follow-up by local team members to full electronic medical records. Telehealth platforms offer an innovative approach to bridging diagnostic gaps by connecting patients in remote regions with specialists.^27,28

The next step is deciding how to deliver care. Options include recruiting international SLPs to travel and provide direct care or having international SLPs train local SLPs in the diagnosis and management of VPI, following the diagonal model of care. ²⁹ This approach balances short-term improvements in VPI management capacity with broader efforts to strengthen local healthcare systems. ²⁹ The success of these strategies requires the willingness of international providers to dedicate time away from their practices, the ability of host healthcare systems to incorporate VPI care into their frameworks, and the adherence to compensation and quality assurance standards to ensure sustainability. ³⁰ Even with successful implementation, establishing a robust care framework will require years of coordinated effort.

Artificial intelligence has the potential to address many of the limitations in current VPI care models. Artificial intelligence–powered diagnostic systems can analyze speech patterns to detect hypernasality and other VPI markers with high accuracy. For instance, validated models such as the Objective Hypernasality Measure and Cross-Attention Residual Siamese Network have demonstrated remarkable precision in differentiating VPI from normal speech.^31,32 These scalable, objective solutions can be seamlessly integrated into telehealth platforms. By leveraging AI-powered mobile applications, healthcare providers can efficiently triage patients who would benefit most from evaluation by SLPs, streamlining diagnostic workflows. This targeted approach can optimize resource allocation and reduce the financial burden of care by minimizing false positives and unnecessary referrals. Additionally, AI-powered diagnostic systems can be integrated into smartphones, creating a low-cost, portable triaging device. ³³ With thoughtful validation and integration into clinical workflows, AI-driven technologies hold the potential to bridge critical gaps in VPI diagnosis and treatment, particularly in underserved and resource-constrained settings. ³⁴

The promise of AI is not without challenges, as its implementation must be both technically feasible and contextually appropriate. Developing AI-powered diagnostic tools for global use requires addressing fundamental barriers related to data collection, infrastructure, and cultural nuances. Challenges include differences in privacy laws, variability in data collections, and infrastructure limitations, such as unreliable Internet access in many LMICs. Overcoming these barriers will require international collaboration to establish secure data-sharing agreements and standardized frameworks for data collection and use. Additionally, AI point-of-care systems must account for cultural and linguistic differences in speech patterns, incorporating multilingual support and localized calibration to ensure accuracy across heterogenous populations. Furthermore, AI systems must be continuously monitored and refined to ensure fairness, generalizability, and reliability over time. ³⁵ Without regular updates and oversight, these systems risk perpetuating biases or becoming less effective as they are applied to diverse populations. ³⁵ By addressing these challenges, AI-powered diagnostic tools could amplify the capacity of existing SLPs without requiring substantial changes in local healthcare infrastructure.

Focusing on future steps beyond primary cleft lip and palate repair is important for bringing comprehensive cleft care to all patients worldwide. The development of intelligible speech is critical for the integration of children into society, maintaining meaningful relationships, and contributing to the local workforce and economy. Tracking speech outcomes in LMICs will help identify the prevalence and severity of VPI, uncover gaps in care, and generate data to advocate for better resource allocation. These efforts can inform strategies to scale up access to VPI care effectively. Artificial intelligence and machine learning offer scalable, innovative solutions to achieve these goals without necessitating significant changes to existing healthcare infrastructure.