Augmented Cognition – The Augnito Blog

Augmented Cognition

The Augnito Blog

The role of
speech recognition
in radiology security, risk and compliance

31 Jul, 2022

Adopting any new technology is a cautious process for many, balancing the wellbeing of radiologists with the needs of patients and concerns over privacy and risk. However, modern, next-gen speech recognition not only delivers compliance, but can de-risk the way you already create and manage reports.

Driven by Government expectations and its strong recommendation to the public sector, hospital Trusts are pressured to replace legacy systems and adopt new technology that demonstrates better value for money. The Cloud First policy is just one example: a comprehensive initiative to increase cloud adoption throughout the NHS.

Radiologists tend to explore new technology with confidence – they were one of the first to adopt speech recognition. However across entire departments and frontline healthcare organisations, several barriers can make more widespread adoption difficult. According to Matthew Chase, former CTO at Guy’s and St Thomas’ NHS Foundation Trust, these include concerns over cost and complexity – but security is an issue that’s raised most often.

Security in particular holds Trusts back from innovations that could empower radiologists, improve collaboration, and ultimately improve patient care.

More than anything else, radiology reports are valuable sources of data and insights that will inform patient management. Protecting that data is a critical task.

Creating consistent, compliant reports

Traditionally, the narrative reports from different radiologists vary in language, length and style. As a result, they’re not always clear when shared with referring clinicians, adding a layer of translation and decryption to an already difficult workload.

Increasingly, technology is used to bring consistency and control to the way radiologists report. This significantly improves reporting quality – but it also means entrusting sensitive patient data to the platforms you use as a department.

In part, effective security is about protecting this data from leakage, loss and malicious interception. But it’s also about safeguarding the integrity of the valuable report data created using years of radiology experience and learning, particularly when it’s used to determine an overall path of care.

The three security barriers for speech recognition

Accurate speech recognition has the potential to further transform radiology as a field and transform the lives of radiologists. However, until now, adoption has been limited by the understandable concerns of Trusts that don’t want to be exposed to additional risk.

#1 The transfer of data to the cloud

The cloud and electronic health records are already making it easier for clinicians to share information, monitor cases at every stage, and understand every patient interaction in context. Despite this significant benefit, many are concerned about the safety of data as it is transferred outside of the local infrastructure.

More than servers and connections, the cloud is built on trust. It’s a major consideration in cloud adoption, with users expecting guarantees on:

  • How data is made accessible to specific individuals
  • How data is secured against unauthorised access
  • How data is protected against tampering or manipulation
  • Availability, resilience and low or no risk

The most effective, accurate speech recognition uses cloud computing to support complex AI models. Robust data protection and security is vital.

#2 The use of data to improve AI

Speech recognition doesn’t just absorb voice data and create reports. Any machine learning AI will need to process and learn from data, contributing towards natural language processing and natural language understanding. As standard, this data should always be anonymized and encrypted before it is used in training.

In this context, the integrity and security of voice data is a concern that should stretch beyond radiology or individual Trusts. A speech recognition provider can only deliver accuracy that continually learns and improves by embedding robust security and data protection – removing any identifiable data, or risk of data breach and loss.

#3 Maintaining compliance at every touchpoint

With the rise of remote working, governance has become a big challenge for healthcare. Compliance with regulatory expectations is no longer confined to the hospital and its core systems. It’s something that needs to go wherever users do – and wherever data does.

Native speech recognition on smartphones and tablets creates a complex compliance minefield. When using solutions like Siri, for example, do you, the user, know where the audio is being routed to for processing? Do you know what data/personal detail is being stored and where? Could there be a risk of cross-border routing in a way that’s likely to breach GDPR?

Overcoming this compliance concern takes speech recognition that’s expressly built for the needs of healthcare and radiology, with an acute focus on keeping data safe, secure and compliant.

How next-gen speech recognition keeps data secure

Developed in partnership with clinicians, for clinicians, Augnito’s AI-driven speech recognition and voice transcription offers industry leading security. It’s not just a dictation platform for accurate transcription, but a powerful tool that radiologists and Trusts can feel comfortable about using.

A few of Augnito’s parent company’s security and compliance credentials include:

  • ISO 27001:2018 certification and GDPR compliance
  • A speech recognition engine hosted by Amazon Web Services (AWS) in the UK
  • Anonymised, encrypted data at rest, store exclusively in the UK
  • Automated backup options in AWS to prevent data loss

In these ways, Augnito goes beyond accuracy to offer security, data protection and peace of mind.

At the same time, Augnito doesn’t just take the risk out of adopting speech recognition – it’s a powerful way to raise your security standards in the cloud compared to traditional, manually typed reports, and offers value for money compared to legacy, or non-clinical voice recognition solutions.

By bringing consistency and reproducibility to every radiological report, enabling role-based, secure sharing, and enabling secure transcription on any device, Augnito is making radiology data safer than ever before.

Related posts

The costs of risk and errors in radiology reporting

The costs of risk and errors in radiology reporting

While next generation speech recognition has the potential to further transform radiology, no change comes without some degree of cost and complexity. But with rising backlogs, a culture of litigation, and growing risk for radiologists in the UK, the costs of doing nothing could be significantly greater.

read more