Advancing Conversational AI in Healthcare
At Hana, we're pioneering research into how conversational AI can transform patient engagement and care delivery.
The Adaptive Engagement Engine
Our proprietary multi-model intelligence system translates research into practice by continuously monitoring conversations, analyzing vocal biomarkers, and adapting engagement in real time.
Core Capabilities
Publications
Discover the research and clinical evidence foundational to our products
Voice in Parkinson's disease: a machine-learning study
Machine learning models detect Parkinson's-related voice changes even in early stages. Using 115 patients and 108 controls, algorithms differentiated early vs. mid-advanced PD and quantified therapy effects. L-dopa improved but did not fully restore vocal quality. Demonstrates how non-invasive voice monitoring allows continuous tracking and personalized care for patients.
Artificial-intelligence-based voice assessment of patients with Parkinson's disease off and on treatment: machine vs. deep-learning comparison
Compared traditional ML (k-NN, SVM, Naive Bayes) and deep learning (CNN) approaches for PD voice assessment. Both achieved comparable accuracy, with CNN slightly ahead, across vocal parameters including jitter, shimmer, and harmonic-to-noise ratio. AI-powered voice monitoring enables remote feedback loops between patients and clinicians, enhancing engagement.
Voice biomarkers for detecting depression
Acoustic markers (pitch variability, tempo, pauses) correlate with depression severity. CNN and LSTM achieved up to 90% accuracy. Enables continuous mental health monitoring through voice AI.
Conversational agents in healthcare: scoping review
Across 17 studies, voice-enabled agents improved adherence and self-management in chronic conditions. Noted design gaps: patient trust and engagement.
Voice biomarkers predict hospital readmission in heart failure
ML models trained on HF patients achieved AUC 0.80 for readmission prediction. Voice AI monitoring supports proactive outreach and early intervention.
Voice-based conversational agents for older adults
Virtual health counselors maintained weekly interaction for 6 weeks, improving adherence and mood. Shows how medical voice agents enable patient engagement among elderly users.
AI and vocal biomarkers of cognitive decline
Recurrent neural networks identified mild cognitive impairment with >80% accuracy. Voice as an early screening tool for neurodegenerative diseases.
Patient engagement through voice-enabled virtual assistants
Meta-based chronic disease evidence-based medication adherence by 19%. Validates voice AI as a home-based adherence enabler.
Detecting anxiety via voice and speech patterns
CNN-RNN hybrid achieved >85% accuracy in identifying anxiety states. Voice AI can flag distress for adaptive support.
The promise of voice in healthcare: ambient clinical intelligence and patient experience
Reviews how voice AI streamlines documentation and boosts near-time empathy signals. Voice as the core engagement tool, unifying patients and providers.