In the healthcare sector the areas of application are steadily increasing that we can predict savings by 2026 – according to a recent Accenture research – of 150 billion dollars a year to a complex healthcare system like that of the United States. The estimates say that the potential improvement of care is around 30-40%, with a reduction cost of 50%.

Other analysis underlined – for the period 2019-2025 – an annual growth trend of AI in healthcare between 43 and 50% in a market that already reaches 2 billion dollars; for more information read the report “Driving the future of health” di Strategy&/PwC.

The engine of this global scenario are tech giants such as Google, Amazon, IBM, Alibaba, Intel, that are able to developing big AI projects, for example DeepMind Health (Google) that can manage in a few minutes millions medical data for clinical or diagnostics activities, or Baseline Study for the collection of genetic data; or Watson (IBM) applied to hospitals, able to anticipate in two years heart failure diagnosis for each patient, thanks to the analysis and combination of data of other diseases, prescribed drugs and medical records of previous hospitalizations.

The solutions for healthcare based on Artificial Intelligence, NLP (Natural Language Processing, comprehension of natural language), Computer Vision and Deep Learning (the research area on AI machine learning developed on neural networks) are about all application sectors – not only diagnostics – which contribute to maintaining the people good health.

Major applications aimed to have more precise diagnosis, improve preventive medicine, personalized cares, optimize and simplify clinical processes and planning procedures, manage big non-homogenous data set for predictive analysis, simplifying bureaucratic and repetitive activities managed by doctors and reducing final health costs.

We therefore talk about algorithmic applications in genetics, epidemiology, molecular biology, radiology, Clinical study management, biomedical data use for the management of chronic or rare disease, but also pharmacovigilance, with the prediction, for example, of possible drug side effects, based on the analysis of their chemical structure (cfr. Stanford University study on ACS Central Science magazine).

The imaging is certainly one of the main sector of AI in healthcare, that is why diagnostics for images has made big step forward towards more effective results and more personalized cares: the algorithm of deep learning of Stanford University is a perfect example, refined for an accurate diagnosis of different types of skin cancer through pictures taken by smartphone.

Flash from the Technical Workshop:

The scenario described has been addressed with a focus on AI and diagnostics during one of the most interesting technical workshops to which JSB R&D staff have attended, as a step forward for updating and empowerment.

Here some pills from the event:

Utility of the algorithm

AI is the basic technology on which the algorithm is based. This mathematical function, that is speed and able to analyse and manage loads of data in a short time, is used for:

  1. Standardization of guidelines in clinical practice, supporting patients and medical decisions.
  2. After learning through the ‘deep neural netwok’ (DNNs), that is a sub discipline of ML that find the correct Mathematical “manipulation” to transform input into output, whether in linear or non-linear relationship.
  3. Natural Language Processing

Specific applications of AI in prevention medicine, diagnosis and treatments:

  1. Radiology: accuracy of 0.76 in the forecast of ‘Pneumonia’ bacteria in 112.000 images
  2. Dermatology: screening of moles to detect melanoma
  3. Genetics: biome sequencing
  4. Prediction of schizophrenia and psychosis

In general, the AI application in healthcare can improve the workflow management, reducing potential errors, speeding diagnosis, identifying care model and assist patients in processing their data independently

From AI algorithm to clinical practice:


AI Limits in helthcare:

  1. The validation of algorithm performance is not equivalent to clinical efficacy
  2. There is a need of more results to improve routine clinical data
  3. Bias on data input

The future of AI? Quantum 2.0

Quantum technology translates some of the properties of quantum mechanics into practical applications, quantum superposition, quantum entanglement and will be finalized by 2025 while quantum computers will be on the market starting from 2030. Quantum computing have a first experimental application in some industrial sectors, such as the one of personalized drugs. Even in this field is emerging the work of Google, that have launch a Quantum Artificial Intelligence Lab for Quantum R&D and AI.

The main applications of this futuristic technology are:

  1. Cloud computing
  2. 5G Communications
  3. Sensing technology
  4. Cryptography

In the future, the Quantum technology will be disruptive in many business sectors, such as Life Sciences and open new market frontiers.

Speakers of the event:

  • Artificial Intelligence (AI) in Health // Karl A. Stroetmann, Empirica Communication & Technology Research
  • AI & Machine Learning algorithms – integrating them to everyday work // Markus Karileet, Solutions Architect, HELMES
  • Real life applications of AI – Healthcare // Sachin Gaur, Director, InnovatioCuris