152 - APHASAI VOICEBRIDGE NA AFASIA ADQUIRIDA ASSOCIADA A TUMORES CEREBRAIS E À RESSECÇÃO NEUROCIRÚRGICA

Doi: 10.47094/978-65-284-0495-7/1352-1363

PALAVRAS CHAVE: Neuro-oncologia. Tecnologia assistiva. Linguagem funcional.

KEYWORDS: Neuro-oncology. Assistive technology. Functional communication.

ABSTRACT:
Brain tumours involving the language-dominant hemisphere and neurosurgical resection within eloquent networks may cause acquired aphasia, including anomia, phonemic or semantic paraphasias, impaired spontaneous discourse and perioperative functional fluctuations. This chapter describes the neuro-oncological adaptation of Aphas AI Voice Bridge, a multimodal web-based communication support platform, and analyses an initial experience involving ten volunteer families. This applied, exploratory-descriptive mixed-method study was framed as a technological experience report. The prototype combined a clinical-contextual profile, personalised phrase training, automatic speech recognition, convolutional neural network analysis of spectrograms, contextual semantic inference, Bayesian uncertainty estimation, continuous listening and human validation. Family reports were descriptively analysed across three predefined domains: communicative assertiveness and inclusion, mood, and willingness to engage in rehabilitation. Nine families, representing 90% of the cohort and 100% of available final assessments, reported markedly more assertive and inclusive communication, improved mood, and greater willingness to participate in therapy. The tenth final assessment was unavailable because the patient died during follow-up. These findings indicate acceptability and perceived usefulness rather than clinical efficacy. Prospective validation using standardised language, functional, usability, safety and uncertainty-calibration outcomes is required before clinical deployment.

Autores

• Bruno Ricciardi Pereira

• Leonardo Ricciardi

• Thelma de Barros Machado