Exploring Speech Language Pathologist’s Awareness and Attitude Towards Artificial Intelligence in Dysphagia

INTRODUCTION

Dysphagia, or impaired swallowing function, refers to the challenge of transporting a food or liquid bolus from the mouth to the stomach.^[1] Persons with dysphagia have trouble swallowing food and liquids, which could have adverse health and psychosocial consequences.^[2] This can make eating particularly challenging. As a result, dysphagia often hinders the intake of sufficient calories and fluids, which can lead to serious health issues. Literature reports that dysphagia has an adverse effect on one’s quality of life (QoL), causing negative physical implications such as malnutrition, dehydration, aspiration pneumonia, and psychological and social problems.^[3] Dysphagia is becoming increasingly prevalent, with a reported 47.71% prevalence in India,^[4] and 34.16% of older adults in India have self-reported swallowing difficulties, with 45.19% reporting a poor QoL related to these issues.

Speech language pathologists (SLPs) use various techniques, such as clinical bedside evaluations and instrumental assessments like the modified barium swallow study (MBSS) and fiberoptic endoscopic evaluation of swallowing (FEES), to assess swallowing function.^[5] After assessing the nature and severity of dysphagia, an SLP may either recommend no oral intake if feeding and swallowing are unsafe or suggest modified methods to enable safe ingestion.^[5,6] SLPs assess swallowing and make recommendations for safe oral intake, diet modifications, positioning, and swallowing exercises. They may also refer patients for further evaluations, such as gastrointestinal or ENT assessments. In acute care, SLPs focus on establishing a safe diet, working closely with nursing staff to reinforce techniques such as double swallowing, alternating solids and liquids, using proper head postures, and ensuring adequate assistance during meals. They also ensure medications are taken in the appropriate form for safe swallowing.^[7] A treatment plan should include recovery, compensatory, and adaptation strategies, all applied systematically by a qualified professional to achieve the best rehabilitative outcome.

The term “artificial intelligence (AI)” is a fusion of “artificial,” signifying man-made or non-natural, and “intelligence,” denoting the capacity to acquire, process, and apply information for adaptive responses within a given context.^[8] AI represents a field within computer science where computers emulate human-like behaviours and cognitive abilities, enabling them to execute tasks.^[9] The objective of AI is to construct machines capable of exhibiting intelligent behaviours.^[10] AI applications have revolutionised healthcare, making patient monitoring and management through virtual care and advanced wearable technology a standard practice. AI has also introduced innovative solutions in rehabilitation, encompassing both physical (robotics) and virtual (informatics) aspects. For instance, smart mobile and wearable devices can collect data and offer users insights to assess health improvements and track progress toward personalised rehabilitation goals.

AI is revolutionising dysphagia assessment and management by enhancing diagnostic precision and streamlining therapeutic decision-making. Deep learning models, particularly convolutional neural networks (CNNs), have demonstrated high accuracy in interpreting Video Fluoroscopic Swallow Studies (VFSS), offering detailed insights into swallow physiology and aiding individualised treatment planning.^[11] AI-powered video analysis tools, such as those utilising the YOLOv7 model, enable remote, efficient VFSS interpretation, especially beneficial for patients with mobility challenges.^[12] Additionally, machine learning algorithms like Random Forest classifiers have shown promise in non-invasive dysphagia screening through cervical auscultation, achieving up to 83% accuracy in distinguishing normal from pathological swallowing sounds.^[13]

Numerous AI-based applications, such as SWALLO, Gokuri, and Swallowscope, are available for the assessment and management of dysphagia. The “Gokuri” app offers guided swallowing exercises, progress tracking, and educational resources to facilitate effective rehabilitation.^[14] Similarly, the Swallow Scope app aids in managing symptoms with therapeutic exercises and educational content and connects patients with healthcare providers.^[6] The SWALLO Dysphagia App helps in assessing, managing, and tracking dysphagia with standardised assessments and treatment recommendations. SWALLO is already available and shows promise for broader use. Gokuri is currently used in clinical settings and may become more accessible, while Swallowscope represents a promising tool under active research. AI-assisted tools, such as the AIMS-DO app, enhance early detection and diagnostic accuracy by analysing clinical data and symptoms, aiming to improve patient outcomes through personalised and efficient care.

Exploring the knowledge and acceptance of AI tools in the practice of SLPs is crucial due to their pivotal role in the evaluation, diagnosis, and management of dysphagia. With such responsibilities, SLPs are uniquely positioned to benefit from advancements in technology, including AI applications that could enhance the precision and efficiency of their assessments and interventions. Given that SLPs work collaboratively with other healthcare professionals, including dietitians, occupational therapists, and physicians, AI tools could enhance interdisciplinary communication by centralising data and improving workflow efficiency. However, for AI tools to be integrated effectively, it is essential to assess SLPs’ current knowledge, attitudes, and readiness to adopt these technologies. Researching their acceptance of AI could help identify barriers, such as a lack of training or uncertainty about AI’s benefits, and provide valuable insights into how these tools can be better incorporated into clinical practice.^[14,15]

Currently, no study has specifically explored the awareness and attitudes of SLPs towards AI-based applications for dysphagia intervention in India. Understanding their perspectives on integrating AI into daily practice is essential for identifying barriers to adoption, such as a lack of training or uncertainty about AI’s clinical benefits.^[14,15] Therefore, this study seeks to assess the knowledge, perceptions, and attitudes of SLPs toward AI tools and their readiness to embrace technological advancements in dysphagia management.

MATERIAL AND METHODS

A cross-sectional research design was employed, using convenience sampling to gather responses from practicing SLPs across India between March and August 2024. The sample size of 97 participants was determined based on the reported prevalence of dysphagia in India ¹⁶ and calculated using the standard formula n = Z²pq/e², z = 1.96, p = 50%, q = 50%, and e = 0.1, where z is the z-score, e represents the margin of error, n is the population size, and p refers to the estimated population proportion. Convenience sampling was selected to enable participation from professionals working in varied clinical and academic settings. Although this method may introduce selection bias, deliberate efforts were made to include participants from diverse work environments to ensure broader representation.

The study was approved by the departmental ethics committee of Bharati Vidyapeeth (Deemed to be University) School of Audiology and Speech Language Pathology, Pune, and the ethical guidelines were strictly adhered to throughout the study.

The questionnaire was designed to explore SLPs’ awareness, knowledge, attitudes, and readiness to adopt AI in dysphagia-related practice. The initial version of the questionnaire consisted of 25 items, created following an extensive literature review and informal consultations with domain experts. These questions were intended to cover four domains: (1) demographic details, (2) awareness and knowledge of AI in dysphagia, (3) perception and attitudes toward AI use, and (4) willingness to implement AI-based tools in practice [Supplementary S1].

Content validation was carried out by three experienced SLPs with expertise in dysphagia. Each item was assessed for relevance and clarity using a 5-point Likert scale ranging from 1 (unsatisfactory) to 5 (excellent). Content validity index (CVI) was computed at both item and scale levels. The item-level CVI (I-CVI) was calculated as the proportion of experts rating each item ≥3 (i.e., satisfactory or higher), and the scale-level CVI (S-CVI) was derived by averaging the I-CVI values. Items that did not meet the minimum acceptable I-CVI threshold of 0.78 were either revised or removed. Following expert review, three items were discarded due to redundancy, resulting in a refined set of 22 items. The final I-CVI and S-CVI scores were 0.84 and 0.87, respectively, indicating strong agreement among raters.

To assess reliability and internal consistency, Cronbach’s alpha was calculated for the final questionnaire, yielding a value of 0.767, which reflects good consistency. One item from the “willingness to incorporate AI” domain was reassigned to the “perception and attitude” section based on expert suggestions to improve conceptual alignment.

A pilot test was conducted with a sample of 10 SLPs not included in the main study. Participants reviewed the questionnaire for clarity, ease of understanding, and time efficiency. Feedback from the pilot testing led to minor wording adjustments and improved response options, ensuring the questionnaire was user-friendly and comprehensible.

The finalised 22-item questionnaire comprised four sections: Section 1 (demographics) with eight questions addressing aspects such as age, gender, qualifications, experience, and work setting; Section 2 (awareness and knowledge) with six questions, including one open-ended item; Section 3 (perception and attitude) with six questions, including two open-ended and one multiple-choice question; and Section 4 (willingness to incorporate AI) with two questions, one of which was open-ended.

Data collection was carried out through Google Forms, which were distributed online to SLPs working in a variety of settings, such as clinics, hospitals, and institutes in India. The inclusion criteria required participants to have at least a bachelor’s degree in Audiology and Speech-Language Pathology (BASLP) from a university recognised by the University Grants Commission (UGC) of India and accredited by the Rehabilitation Council of India (RCI). The exclusion criteria apply to SLPs with less than 3 months of experience.

Informed consent was obtained from each participant prior to their involvement in the study, and they were also informed that their participation in the study was entirely voluntary. If they agree to take part but change their mind, they are informed that they have the right to do so. They can withdraw from the study at any time. Participants were also provided with sufficient details about the study’s aim and objectives, allowing them to make an informed decision about their participation. This ensured transparency and respect for their autonomy throughout the research process.

Data Analysis: Responses were analysed using descriptive statistics to summarise the demographic information and responses to the knowledge, awareness, and attitude questions. Inferential statistics, the Chi-square test was used to examine associations between demographic variables (e.g., qualification, specialisation, years of experience, workplace) and awareness, attitude, perception, and willingness of AI. Statistical analyses were performed using SPSS (Statistical Package for the Social Sciences) version 25.0 software (IBM Corp., 2012 release), with a significance value of p < 0.05.

RESULTS

DISCUSSION

The findings of this study highlight a considerable gap in awareness and usage of AI tools in dysphagia management among Indian SLPs. Only 20.6% of respondents were aware of such applications, and an even smaller proportion (16.5%) reported using them in clinical practice. Studies support this finding that SLPs working globally report a moderate understanding of AI tools, especially generative technologies like ChatGPT, due to limited awareness.^[17] These low figures suggest limited integration of AI in speech-language pathology, despite its expanding role in healthcare. AI has been widely recognised for its ability to process complex datasets and offer insights in diagnostics, patient care, and decision-making.^[17] However, its adoption in dysphagia management remains slow, possibly due to unfamiliarity with relevant tools, limited formal training, and a strong reliance on traditional practices. Other literature also shows that awareness and adoption of these tools are influenced by the perceived benefits of AI, including enhanced efficiency and improved job satisfaction.^[18]

Notably, most participants acquired knowledge of AI informally, through the internet or peer discussions, rather than through structured educational programs. This aligns with prior research, which suggests that SLPs often encounter AI through personal exploration, especially due to the growing public attention around generative AI tools such as ChatGPT and DALL·E 2. The present study found that only 2.1% of respondents had received formal training on using AI for swallowing assessments, underscoring the need for structured learning opportunities and capacity-building initiatives in this area.^[19]

Another important consideration is that only 14.3% of the participants reported completing additional training in dysphagia. This limited representation of formally trained clinicians presents a notable limitation when interpreting findings related to the use, knowledge, and attitudes toward dysphagia management tools, particularly those involving AI. The relatively small subgroup with advanced training may influence the depth and reliability of responses, especially on questions requiring a nuanced understanding of clinical dysphagia assessment and intervention. Consequently, some of the perceptions and reported practices may reflect limited hands-on experience, which should be factored in when generalising these results to the broader SLP population. Given this, the reliability of responses concerning the use and perceptions of dysphagia management applications may be affected, as participants with limited dysphagia expertise might provide less detailed or accurate insights.

The study also revealed a significant association between participants’ qualifications and their perceptions of AI tools, particularly in relation to the clarity and relevance of the information these tools provide. This suggests that higher educational attainment may be linked to greater openness and confidence in adopting AI. Postgraduate SLPs may have more exposure to interdisciplinary knowledge, research methodologies, and innovative practices, which could positively influence their perception and willingness to adopt AI. On the other hand, BASLP-level practitioners may have limited training or exposure to technological tools, potentially leading to hesitancy or lower confidence in their use.

While more than half of the respondents (52%) recognised AI as beneficial for dysphagia therapy, citing advantages such as time efficiency, ease of use, diagnostic support, improved accessibility, and better patient engagement 43% remained uncertain about its effectiveness. This uncertainty reflects a lack of direct experience and awareness, suggesting that many clinicians have not yet engaged deeply with AI in their practice. Furthermore, 33 participants held no strong opinion on incorporating AI, which further illustrates the early stage of AI adoption among SLPs. This finding echoes prior studies, which report skepticism among healthcare professionals about the reliability of AI, especially when compared to hands-on clinical experience.^[20]

Encouragingly, 72.2% of SLPs expressed interest in gaining skills related to AI usage, and 55.7% believed that its use will expand in speech-language pathology in the future. These attitudes indicate a readiness to embrace technology, provided that adequate training and support mechanisms are in place. Many participants viewed AI as a complementary tool that can enhance traditional therapy by improving diagnostics, reducing clinician workload, and ensuring consistent follow-ups, especially in rural or underserved areas. This finding is consistent with the literature supporting the use of AI tools, such as SWALLO and Swallow Scope, which have shown promise in enabling remote patient monitoring and follow-up in dysphagia care.^[19]

Despite the positive outlook, several barriers to AI adoption were reported. These included lack of knowledge and expertise (63.9%), insufficient access to technical equipment (50.5%), ethical and privacy concerns (24.7%), complexity of AI tools (23.7%), and lack of training (54.6%). The perceived lack of acceptance among professionals (25.8%) also emerged as a hindrance. These challenges highlight the need for tailored workshops, inclusion of AI modules in academic curricula, and clinician-friendly tool designs to facilitate seamless integration into practice.

AI holds promise for improving diagnostic accuracy, personalising treatment, and enhancing outcomes, particularly in settings with limited resources. However, the present study underscores the need for systematic education, greater awareness, and accessible infrastructure to support the implementation of AI in dysphagia management. While AI may not replace the clinician’s role, its collaborative use in clinical decision-making can contribute to improved care and QoL.^[21]

CONCLUSION

In conclusion, while AI integration in dysphagia management in India remains in its early stages, this study reveals a growing awareness and cautious optimism among SLPs, tempered by limited training, clinical exposure, and resource availability. Despite concerns about AI’s current reliability and inability to replace clinical judgment, there is clear potential for it to enhance diagnostic accuracy, support remote care, and reduce clinician burden. However, a major limitation influencing the interpretation of these findings is the use of convenience sampling, which may reduce the generalisability of results and introduce selection bias. Additionally, the unclear proportion of participants actively managing dysphagia cases, combined with the predominance of undergraduate-level professionals, may affect the depth and applicability of the responses. Uneven geographic representation and the relatively brief study period further restrict the ability to capture long-term trends in AI adoption among SLPs. To advance AI adoption, structured educational initiatives and validated training modules are essential, ensuring that SLPs are equipped to integrate AI tools ethically and effectively, while maintaining the critical human element in dysphagia care clinical settings.