Talking to the Implant Boston Scientific’s AI System That Programmes Neurostimulators Using a Patient’s Own Words

Boston Scientific has patented a system that uses natural language processing to interpret a patient’s freeform text or voice descriptions of how they are feeling, then uses that information to evaluate and adjust the programming of an implanted neurostimulation device. The invention brings conversational AI into the clinical management of pain and neurological conditions – making it easier for patients and clinicians to communicate treatment experience and make informed therapy adjustments.

The Problem

Neurostimulation devices – such as spinal cord stimulators, deep brain stimulators and peripheral nerve stimulators – are implanted medical devices that deliver targeted electrical impulses to specific parts of the nervous system to manage chronic pain, movement disorders and other neurological conditions. Programming these devices optimally is both critically important and genuinely difficult.

The challenge is that the therapeutic effect of neurostimulation is highly subjective – it depends on the patient’s own experience of pain relief, side effects and functional changes. Clinicians must rely on patients to accurately describe their experience during programming sessions, which typically happen at scheduled clinic visits. Between visits, patients may experience changes in their therapy – side effects, reduced efficacy, new sensations – but communicating these precisely in clinical language is difficult for many patients. Medical terminology creates a barrier between what the patient is actually experiencing and what gets recorded in a structured clinical assessment.

Meanwhile, the neurostimulation device itself generates data that reflects the state of therapy delivery – but this device data alone cannot capture the patient’s subjective experience. What is needed is a way to bridge the gap between the patient’s own words and the structured data that drives clinical decision-making.

What This Invention Does

Boston Scientific’s system obtains text content from the patient – either typed freeform text or transcribed voice input – that describes the patient’s current state. Natural language processing analyses this text to identify the patient’s state in clinically meaningful terms: their pain level, functional status, side effects experienced and other relevant health information.

Simultaneously, the system retrieves data from the neurostimulation device – information about the therapy being delivered, stimulation parameters and device status. It then associates the patient’s identified state (derived from their words) with the identified state of the neurostimulation treatment (derived from device data), creating a linked picture of how the therapy is performing from the patient’s perspective.

Based on this association, the system can initiate an action – such as flagging the case for clinical review, suggesting a programming adjustment, or routing the patient to appropriate follow-up care. This “triage” function means patients whose language suggests inadequate therapy or problematic side effects can be identified and prioritised without waiting for a scheduled visit.

Key Features

Freeform text and voice input. Patients can describe their experience in their own words – spoken or typed – rather than being constrained to structured questionnaires or numerical scales, lowering the communication barrier and capturing richer, more nuanced patient-reported information.

Natural language processing for state identification. The system applies NLP to interpret the patient’s freeform input and identify their health state in terms that can be correlated with device performance data – bridging the gap between patient language and clinical assessment.

Device data integration. By combining the patient’s reported state with objective data retrieved directly from the neurostimulation device, the system creates a comprehensive picture of therapy performance that neither source could provide alone.

Automated triage capability. The system can initiate appropriate actions based on the identified patient-therapy association – directing urgent cases to faster follow-up, suggesting programming adjustments or providing the patient with relevant guidance.

Continuous between-visit monitoring potential. The freeform text model enables patients to report their experience at any time, not just during clinic visits – supporting more responsive, continuous management of neurostimulation therapy.

Who Is Behind It?

Boston Scientific Neuromodulation Corporation is a division of Boston Scientific Corporation, one of the world’s largest medical device companies. The inventors are Kyle Harish Srivastava, Benjamin Phillip Hahn and Amarpreet Singh Bains. This application is a divisional of AU 2022408043, which entered the national phase from PCT/US2022/051934 filed 6 December 2022, claiming priority to a US provisional application filed 9 December 2021. The application is managed by Spruson & Ferguson in Sydney.

Why It Matters

Neurostimulation is one of the fastest-growing areas in medical device technology, with applications spanning chronic pain, Parkinson’s disease, epilepsy, depression and bladder disorders. As patient populations grow and devices become more sophisticated, the clinical workload of programming management increases. A system that can intelligently interpret patient-reported experience and link it to device data – enabling better triage and faster therapy optimisation – addresses a real bottleneck in neurostimulation care delivery.

The use of natural language processing to interpret freeform patient input also reflects a broader trend in healthcare toward patient-centred outcome measurement. Patients report their experience more accurately and completely when asked in their own words than when constrained to structured scales. With IPC classifications covering health informatics (G16H 20/00) and healthcare treatment monitoring (G16H 20/40, G16H 20/70), the patent sits at the intersection of AI, digital health and implantable device management.

---

AU 2026201524 was published in the Australian Official Journal of Patents on 19 March 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.

Related Concepts

Natural language processing is increasingly being applied to clinical settings, where structured questionnaires and numerical scales often fail to capture the full picture of a patient’s experience. In neuromodulation, where therapy effectiveness is entirely subjective, systems that bridge patient language and device data enable better triage, faster programming adjustments, and more continuous care – reducing reliance on infrequent clinic visits to manage complex, long-term implanted therapies.