AI: The Future of HealthTech

Being bolstered by AI, both automation and hyper-automation have remained leading IT trends for several years in a row while amplifying numerous industries. However, there is one area tech scions never took seriously before the pandemic: the potential of AI in HealthTech.

In 2022, HealthTech has made it into common flagship services and a fast lane across global tech companies. With that said, let’s see how AI can be used in healthcare software, its current stand on the market, and related technologies you might want to look into further. I’ll also go over one of the technicalities of intelligent drug discovery and gene-sequencing, based on real cases from some of my clients.

AI HealthTech Market Is Quickly Expanding

The healthcare applications of artificial intelligence (AI) are growing rapidly, according to a new report from research firm MarketsandMarkets. The company estimates that the global market for AI in healthcare will grow at a compound annual growth rate (CAGR) of 46.2 percent between now and 2027 when AI’s share of the total healthcare IT market is expected to reach $67.4 billion.

Another study by Allied Market Research emphasizes the positive impact of the COVID-19 pandemic on healthcare-related markets, one of them being AI in healthcare. At present, AI technologies are playing a crucial role to combat the pandemic.

HealthTech Trends

Source: Philips

Investment in AI Healthcare Soared and So Did AI Tech Companies

Healthcare organizations vary significantly in their AI investments. Before the pandemic, 75% of large organizations injected over $50 million in intelligent technologies. At the same time, 95% of mid-scale companies put away under $50 million on AI transformation.

In 2021, cash injections into the field sourced. The COVID-19 whiplash resulted in almost 3 in 4 healthcare providers increasing their funding, according to Deloitte. Thus, the total AI investment exceeded $77 billion in 2021, a sharp rise from the previous record set in 2020, which is $36 billion.

As the world continues the pandemic fight, AI’s ability to monitor the spread of COVID-19 cases (94%), assist with vaccine development (90%), and aid vaccine distribution are most favored by life sciences executives (90%).

For a software development world, this automation craze resulted in the AI ​​talent grab.

The Big 5, including Google and Apple, have been bolstering their AI talent through acquisitions. For the record, there have been over 630 AI acquisitions since 2010. Therefore, artificial intelligence along with ML and Deep learning remain among the main software development trends in 2022 due to the increased AI cases in healthcare and other industries.

On this note, let’s have a closer look at the relevant technologies in HealthTech that stem from AI concepts. You might want to tap into these if you’re thinking of joining the HealthTech arena:

Machine Learning: Neural Networks and Deep Learning

These are some of the most common AI forms that are mainly used as predictive tools for precision medicine. Deep learning and neural networks are more complex than supervised ML learning and include many levels of features or variables that project outcomes.

Natural Language Processing

Natural Language Processing is a subfield of Artificial Intelligence that allows computers to comprehend human speech and text. Within medical software, it’s mainly responsible for analyzing unstructured clinical notes, transcribing patient interactions, and implementing conversational AI.

Big data and predictive analytics

These are must-haves to collect continuous data and analyze it at scale. Besides, interoperable systems are now promoting more convenient data sharing for the sake of predictive analytics.

Virtual Reality and Augmented Reality

Healthcare providers adopt VR and AR technologies to enhance their customer experiences by engaging them in healthcare activities, including self-diagnosis and education. Both can also be used for training medical professionals.

Now let’s see how a combination of these technologies can be used to amplify drug research and gene sequencing.

The Key Drug Discovery Player

More importantly, where are these funds going? AI is becoming a more common technique in drug development, design, and target identification for biopharma companies.

Smart technologies have been used in drug discovery research since the 1950s. This practice, known as computer-aided drug design, is the use of computer algorithms and software to predict – from the structure of a chemical compound to its therapeutic potential.

According to MarketsandMarkets, AI-enabled drug discovery accounts for a vibrant market with a global value of $1.4 billion by 2024. This is a significant leap from $259 million in 2019, which can be attributed to the cost-effectiveness and high performance of AI- based drug exploration.

As of today, the cost of bringing a new drug to the market amounts to $2.6 billion. Besides, it takes more than ten years to bring a new drug to the market. Artificial intelligence can significantly reduce costs by automating the process of designing, synthesizing, and evaluating numerous compounds.

Where Am I Headed?

Artificial intelligence and Big data will secure their place as the most bankable skills for software developers and mandate technologies for software companies. Nobody wants outdated ineffective analytical tools with manual input that takes zillion hours.

The year 2021 marked a milestone in AI-empowered drug research. Thus, the German biotechnology company Evotec announced the first stage of clinical trials of a new anticancer molecule. The candidate was developed in collaboration with Exscientia, a UK-based organization that specializes in small-molecule drug research based on AI systems.

Together, biopharma companies successfully discovered a therapeutic candidate in 8 months. This blazingly fast research became possible thanks to Exscientia’s “Centaur Chemist” AI design platform that reduced the traditional discovery timeline from 5 years to less than a year.

Another breakthrough in the field is credited to Google’s subsidiary, DeepMind. The latter has created a comprehensive map of human protein structure using AI software called AlphaFold. Today, the company is rolling out 180000 predictions made by the program to the public. Although it was done before, DeepMind’s database is the most comprehensive and accurate so far.

Two examples of protein structures predicted by AlphaFold (blue) compared with experimental results (green). Image: DeepMind

How Does AI in Drug Discovery Work?

In layman’s terms, Artificial intelligence gives computers the ability to learn without being explicitly programmed. In drug discovery, AI can be used to support or even replace several human tasks, including drug screening and development, lead identification and optimization, target validation, biomarker discovery, and more.

The greatest potential of AI is its ability to process oceans of data in mere seconds. Therefore, smart algorithms can comb through data-laden libraries used to screen for new drug candidates. This then allows researchers to predict the properties of a potential compound, produce ideas for new compounds, and automate mundane tasks.

The step-by-step process of AI-based protein prediction includes the following steps:

  1. Data on a target protein is injected into a database.

  2. The algorithm eliminates potentially dangerous compounds.

  3. The system selects the most promising compounds for further testing.

  4. Other targets are added to the AI ​​system for further selection and testing.

  5. The result of the drug-discovery process is a shortlisted number of potential components for advanced testing.

AI Enabling Genome Sequencing

As a CEO, I have seen first-hand how AI algorithms transform traditional gene sequencing tools into comprehensive real-time analytics solutions that also chip in to fight off diseases. Genes carry information about most characteristics of the human body, including the probability of the gene to suffer mutations and specific diseases.

Gene sequencing and data analysis can help researchers find out more about the genetic makeup and the relationship between diseases and specific gene forms. However, standard nanopore DNA sequence devices (which are available for around 1K) fall short to analyze real-time data. Existing genomic analyzes, on the other hand, include transporting DNA to a centralized facility, sequencing, and analyzing samples in a batch process, which is time-consuming.

That’s when AI algorithms come into play.

The Behind-the-Scenes of AI Gene Sequencing

Let’s see how artificial intelligence is embedded into DNA sequencing exemplified by one of our projects. The project was a real-time DNA sequence analysis application that examines DNA nanopores in real-time with the Google Cloud Platform.

It can detect taxonomic proportions, possible infections and diseases, antibiotic resistance genes, and other things. Here’s how it operates from a tech standpoint:

First, a portable nanopore sequencing device is used as a data source for blood and other DNA samples (specifically, a MinION device). Then, we send the obtained data to the Google Cloud Platform for further analysis.

Step-by-step AI HealthTech example process

The step-by-step process looks like this:

  1. Uploading files to the Google Cloud Platform and streaming them into the processing pipeline (ie ingestion)

  2. The machine learning model infers DNA sequences from electrical signals during the base-calling stage.

  3. The samples are evaluated for pathogen sequences, taxonomic enrichment, and other anomalies using a DNA database.

  4. Calculation of each pathogen’s proportion in the particular sample at the summarization step

  5. The results are saved to Google Firestore DB and then visualized in real-time with D3.js.

Looks like a no-brainer, but a lot is happening in-between stages, including:

  • System architecture (in this case, Google Cloud Platform processing tools were used after studying genomics domain data formats)

  • Machine learning engine (TensorFlow model trained on genome data)

  • Back end (Firebase as a document-storage system friendly to hierarchical data, which is important for representing biological taxonomies)

  • Front end (D3.js dynamic dashboard which updates the Sunburst chart visualization)

This is what a real-time Nanopore DNA sequencer looks like on the outside. As a result, it can generate DNA sequence data from patient samples in mere minutes.

What Does It All Mean for the Tech Universe?

With the rapid advancement of technology and its application in every field, artificial intelligence has become an unavoidable topic in healthcare and mandated expertise for leading tech companies. Along with empowering better clinical decisions and improving patient care, AI is uniquely positioned to accelerate drug discovery to develop an effective treatment for over 10,000 diseases and unravel the mysteries of human genomes. AI-based healthcare applications are already manifold and will continue to flourish calling for more AI talent.

Beyond HealthTech, artificial intelligence is now poised as the leading technology across the software development industry. Thanks to a huge automation potential, AI can be used as a cornerstone for numerous software applications from chatbots to OCR software.

.

Leave a Comment