AI (6)

In 2016, supercomputer IBM Watson diagnosed a rare form of leukaemia in a patient at a hospital linked to Tokyo University. Using Artificial Intelligence (AI) and operating on the cloud, IBM Watson can cross reference and analyse data from millions of international oncology papers. From this data, it can extract information much faster than humans ever can. Evidence of its capabilities were  reported by the University when IBM Watosn correctly diagnosed a Japanese woman in ten minutes.

Health advances initially seen as futuristic, like virtual avatars and chatbots, are quickly becoming a reality. An article in The Irish Times, says these technologies apply AI to match discussions with people, connect to the Internet and perform tasks that normally require human intelligence.

Sensley’s an example. It’s a mobile triage mHealth app currently being tested by the UK’s NHS. Sensley has an AI nurse that guides patients  through their personal healthcare needs. It’s available all day, every day. Dressed in blue NHS scrubs, Seneley collects information by listening and asking questions similar to interactions between a clinicians and patients. Richard Corbridge, the developer and chief executive of eHealth Ireland says “Things are moving so fast that technologies we would have regarded as sci-fi last year, will become a reality this year. Over the last couple of years, Ireland has made some really big strides in digital healthcare.”

Corbridge believes that by 2019, all Irish maternity hospitals will be using advanced monitoring technology for newborns. Every newborn will have three devices in their cot, monitoring respiration, temperature and heart rate. Information will transfer automatically to their EHRs. In Healthcare Dive Corbridge says instead of constantly checking these levels individually, nurses will have a tablet to monitor the vital information and requests for tests, scans and results.“It’s an amazing leap for Ireland in a short space of time,” says Corbridge.

Will Africa’s health systems use AI soon too? Their eHealth strategies should now include a section on medium term plans for adopting AI.

Conventionally, skin cancer’s primarily diagnosed visually. It starts with a clinical screenings, then, if needed, followed by dermoscopic analyses, a biopsies and histopathological examinations. A team mainly from Stanford University, California, has reported in Nature that mHealth can provide an alternative. It’s a technological step up for Africa’s mHealth.

Classifying skin lesions using images is challenging, owing to fine-grained variabilities in their appearance. Convolutional Neural Networks (CNN) offer potential for dealing with fine-grained object categories. The team demonstrates skin lesion classifications using a single CNN, trained end-to-end directly from images using only pixels and disease labels as inputs. Trained CNN used a dataset of 129,450 clinical images and 2,000 skin lesions.

Its performance was tested against 21 dermatologists using proven clinical images from biopsies in two use cases:

  1. Keratinocyte carcinomas versus benign seborrheic keratosis, identifying the most common cancers
  2. Malignant melanomas versus benign nevi, identifying deadliest skin cancer.

CNN achieved performance in both use cases that matched all tested experts. It shows that the algorithms in Artificial Intelligence (AI) can classify skin cancer as well as dermatologists. Equipped with CNN, mHealth can potentially extend dermatologists’ reach beyond their clinics. An impact is lower-cost universal access to vital diagnostic services.

As healthcare researcher teams extend AI across other conditions, it offers Africa’s mHealth initiatives a much wider role and impact. It seems that mHealth can have much more to offer.

As the quest for Artificial Intelligence (AI) expands, Aajoh, a Nigerian start-up based in Lagos, is developing AI for eHealth. A report in Disrupt Africa says Aajoh’s a private beta testing stage. It follows the selection of the firm’s CEO, Simi Adejumo, as one of nine African entrepreneurs for last year’s MITx Global Entrepreneurship Bootcamp, an intensive one week leadership programme for new ventures.

The AI solution aims to let people  input health symptoms using text, audio and photographs to have instant, accurate diagnoses. They can then buy prescribed medication. The eHealth service also includes live consultations with doctors and booking hospital appointments.

The Aajoh app was developed using data from patients about their health symptoms to generate a model. It’s a continuing process to improve the apps’ performance. The aim was to use the purest form of data and help to scale faster than using traditional hospital data.

Aajoh’s already secured partnerships with two hospitals in Nigeria. It has over 32,000 data entries combined. The beta test uses 34 select users from over 500 people on a waiting list. The plan’s to start to expand across Africa by early 2019 towards an ultimate goal of moving healthcare from reactive to predictive, so forecast when people fall ill.

It’s a considerable challenge. If Aajoh’s app’s effective, it can overcome an accuracy deficit in online diagnoses. eHNA reported on research found doctors better at diagnosing than apps. The goal’s there to be scored.

As Artificial Intelligence (AI) becomes more of a talking point in esoteric circles, two organisations working on a project based in Pittsburgh are using it to expand personalised healthcare. Teams from Carnegie Mellon University and the University of Pittsburgh are part of project financed by a University of Pittsburgh Medical Center for six years at between $10 million and $20 million per year. It’s part of the Pittsburgh Health Data Alliance.

Computerworld has a report that the project's taking data from EHRs, diagnostic imaging, prescriptions, genomic profiles, insurance records and wearable devices to create healthcare plans by disease and specific types of people. These personal traits are seen as foundation to develop apps, machine learning tools and services from the project. A smartphone app for people to use to live healthier lives and ward off some diseases and condition’s scheduled for release in about a year. 

How will AI find its way into Africa’s health systems? It’ll need a sustained investment in the data foundation. The WHO eHealth survey revealed that African countries are behind global eHealth profiles, so catching up to break into AI will take a considerable timescale and investment. With the additional benefits that AI can bring, it justifies increased eHealth spending. Financing it’s probably more demanding.

For many health professionals in African countries, access to specialist cardiologist support is not easy. There simply aren’t enough cardiologists to go round. 

Cardiologs helps to address this gap with a web service that supports a critical aspect of assessing heart patients: ECG analysis. It’s based on machine learning algorithms that provide physicians with supportive information for ECG interpretation. This support brings cardiology expertise closer to health professionals to help them manage patients with cardiac disorders. 

Artificial Intelligence (AI) has been developing rapidly. Cardiologs’ offering is a practical implementation presenting a promising opportunity for countries to strengthen ECG analysis and interpretation, without adding additional burdens to already stretched specialist cardiologist resources.

Existing clients include Agoranov, Bpifrance, and medicen Paris Region. eHNA’ll be on the look out for an African implementation, using this type of approach, soon.

There’s no denying that healthcare has undergone dramatic changes in the last ten years. New technology and innovations available to patients enables them monitor and take responsibility for their own health, and improved devices and tools available to doctors and other health professionals can make more informed decisions. Healthcare technology keeps moving along. An article in The Guardian looked at the top eight technologies that’ll keep transforming healthcare. For Africa, the balance and pace of investment in the eight technologies will be different to developed countries.

The smartphone

Although not new, it’s clear that the smartphone’s healthcare potential’s yet to be realised. Smartphones can serve as the hub for new diagnostic and treatment technologies. We’ve seen apps developed to support a wide range of healthcare activities, such as healthier life-styles, diabetic patients, treatment adherence and depression. Patients can also use tools like the AliveCOR ECG, embedded in a smartphone case, which helps interpret heart test results via an app and facilitates sharing with clinicians. They’re also ideal for gathering large amounts of data to improve understanding of diseases in populations.

At-home or portable diagnostics

Clinicians can now bring hospital-level diagnostics devices to patients’ homes, such as portable x-ray machines, blood-testing kits and other technologies.

Implantable drug-delivery

Drug adherence is a big problem, especially for patients with long term conditions. It’s estimated that between a third and a half of all medication prescribed to people with long-term conditions isn’t taken as recommended. Several technologies are already under development to address the problem. There’s sensor technology so small it can be swallowed and combined with drugs in smart pill form. When the pill dissolves in the stomach, the sensor’s activated and transmits data through a wearable patch to a smartphone app. Patients and clinicians can see how well they are adhering to their prescription, though it raises important questions about patients’ privacy and autonomy.

Digital therapy

Digital therapeutics are health or social care interventions delivered using a smartphone or a laptop. They embed clinical practice and therapy into a digital form to provide computerised cognitive behavioural therapy (CBT)

Genome sequencing

Advances in genome sequencing and the associated field of genomics will give doctors a better understanding of how diseases affect different individuals and populations. These genetic profiles of people’s diseases and knowledge of their response to treatment, it should be possible to predict their response to treatment and prognosis more reliably.

Artificial intelligence

Machine learning is a type of artificial intelligence that enables computers to learn without being explicitly programmed, meaning they can teach themselves to change when exposed to new data. Enlitic, IBM’s Watson division and Google’s Deep Mind have started to explore potential applications in healthcare.

Blockchains

Blockchains are decentralised databases that keep records of how data’s created and changed over time. They’re trusted as authoritative records without a single, central authority guaranteeing accuracy and security. Electronic health records are widely used, but they are usually centralised, provided by a small number of suppliers. Some commentators have described how records using blockchain technology would bring benefits like resilience and encourage interoperability, with patients and clinicians given encryption keys to control who sees the data.

Online communities

Social networks bring together people with interests in healthcare to support each other, share learning and provide platforms for tracking health data, helping people manage their condition and contributing to research. 

New technologies bring new opportunities for Africa’s health systems. They can help to improve the accuracy, reliability, availability and add value of information gathered, change how and where care’s delivered and offer new ways to prevent, predict, detect and treat illness. The numerous choices makes rigorous strategies, plans and investment decisions challenging, but essential.