Artificial Intelligence is the emerging technology in pharmaceutical field besides its diverse applications in other fields. It’s gaining popularity in manufacturing of life saving drugs through use of Manufacturing Execution System. As artificial neural network simulates the brain functioning, it can be explored for optimization and delivery of various formulations.
Artificial Intelligence (AI) is a field of computer science that studies how machines imitate the intelligence of their human counterparts. It is involved in building smart machines that are capable to perform tasks which require human intelligence. Besides its multiple approaches, the advancements in the deep learning and machine learning (ML) are creating a model shift in every sector. AI has three levels; first is Artificial Narrow Intelligence known as machine learning which is capable to specialize in one area and one problem at a time; second is Artificial General Intelligence which has a human level of cognitive function. To work properly, it needs to connect with thousands of Artificial Narrow Intelligence systems to imitate human behaviour and the third is Artificial Super Intelligence which can completely surpass any sort of human intelligence. It can make rational decisions, build relationships and decide whether it wanted to be good or evil. The roots of AI can likely be traced back to the 1942 when the American writer Isaac Asimov published his short story ‘Runaround’ though statistical methods for achieving true was discussed in1940s when the Canadian psychologist developed a theory of learning known as Hebbian Learning. It replicates the process of neurons in the human brain, and this led to the creation of research on Artificial Neural Networks; while between the period of 1964 and 1966, ELIZA a famous computer program was created by Joseph Weizenbaum. The program was a natural language processing tool, it can simulate a conversation with a human. Further artificial neural networks made a comeback in the form of Deep Learning in 2015 through AlphaGo program developed by Google. Recently, All India Radio with Vigyan Prasar launched a brand new radio science series that will focus on Artificial Intelligence (AI). The series can be produced in 19 other Indian languages which broadcasted across 121 AIR stations hence is expected to cover over 85 per cent of India’s geographical area.
AI IN VARIOUS FIELDS
AI is used nowadays in various fields such as use of chat boats to communicate with the customers through messaging apps, speech-based assistants, to defend cyber-attacks, for recruiting candidates for job through Pymetrics software etc. Also it has also penetrated into the health care sector and major applications of it include monitoring medicine, treating chronic illness, diagnosing diseases, and surgery support. E.g. IBM has developed a supercomputer and named it Watson, which is a combination of AI and sophisticated analytical software designed basically to answer questions to assist oncologists in taking better decisions for the treatment of cancer. AI also helps to predict risky patients example in case of COVID-19 pandemic and can control the infection on time while robotics can be used in surgery.
APPLICATIONS OF AI IN PHARMA
1. Drug Design and Drug Discovery Process:
Pharma companies across the globe uses machine learning (ML) algorithms to smoothen the drug innovation process. It detects complex patterns in large datasets and is used to resolve problems associated with complex biological networks. Starting from making small molecules further to determining novel biological targets, AI plays a prominent role in drug target identification and validation, phenotypic, target oriented, and as multi-target drug innovation, and for biomarker identification. This minimizes time required for approval of drug and leads to reduction in the cost which can be passed on the customer. For example, researchers can identify and verify novel cancer drug targets using data such as longitudinal Electronic Medical Records (EMRs).
There are various other tools for drug discovery such as ‘DeepChem’ which uses a python-based AI system to find suitable candidates in drug discovery; ‘DeepNeuralNetQSAR’, is also a python-based system driven by computational tools that aids in detection of the molecular activity of compounds; ‘ORGANIC’ which is a molecular generation tool that helps to create molecules with desired properties; ‘PotentialNet’ uses neural networks to predict binding affinity of ligands; ‘DeltaVina’ is a scoring function for rescoring drug–ligand binding affinity; Neural graph fingerprint predicts properties of novel molecules; ‘AlphaFold’ that predicts 3D structures of proteins; and ‘Chemputer’, which helps to report procedure for chemical synthesis in standardized format. The recent ‘DeepBAR’ technique can quickly calculate binding affinities between drugs and targets. It combines traditional chemistry calculations with recent advances in machine learning. It calculates the binding free energy exactly.
2. Epidemic Projection
Artificial intelligence and machine learning models are particularly advantageous for economies that lack medical infrastructure and financial framework to file the spread of infection. A well-known example of this is the ML-based malaria outbreak prediction model, which works as a warning tool for malaria outbreaks and helps the health care providers to take the best action to combat it.
3. AI for Finding Better Cures
Many pharmaceutical companies use AI-based technology to develop drug improvements and find quicker ways to treat various diseases. Presently, companies are using AI to collect and group body compounds before the experts can use them for further insights. The research teams at Novartis, for example, use perceptive images using machine learning to anticipate whether or not untested compounds are worth testing. Hence, machines help determine a range of data sets to create new drugs. Micro-fabrication technology for production of implantable microchips appears to be a promising approach for controlling delivery of drugs. These are capable of opening on command and provide continuous or pulsatile drug delivery. To achieve such controls Fuzzy logic and Neural networks can be applied. AI tools enable the prediction of pharmacokinetics of novel therapeutics. Artificial Neural Networks (ANN) are also able to create nonlinear input-output mappings, analyze the multivariate nonlinear relationships in pharmaceutical research, design the pre-formulations, and predict the behaviour of drugs genetic algorithms.
4. Using AI to Treat Rare Diseases
ANN exhibits its wide applications in proteomics, genomics, data modeling, development of the pharmaceutical products, and prediction of the bioavailability and behaviour of drugs, it have been suggested as tools for capturing cause-and-effect relationships and predicts in-vitro and in-vivo correlations. Healx is a biotech company that mainly focuses on patient care and works to create effective AI-powered platforms. The company has a ‘Rare Treatment Accelerator’ programme, which is an impressive collaboration that uses the concealed power of AI and investigates better treatments for any rare disease. The company created platform to motivate scientists to enhance their production in rare disease discovery. Instead of using AI to slow down the risk of rare diseases, the company focuses on analyzing drugs and repurposing them to cure rare diseases.
5. Using Image Recognition for Drug Adherence
AI has supported ‘AiCure’ which is advanced data analytics company to create an image recognition platform that prevents the need to measure drug adherence. The amazing facial recognition software is built that helps to trace the consequences of various medication when used by patients. According to recent research, using AI has sharply increased adherence in schizophrenic patients.
6. Prediction of Bioactivity:
As activity of drug molecules depends on the ability to bind receptors, drug target binding affinity (DTBA) is vital to predict drug–target interactions. AI-based methods can measure the binding affinity of a drug by considering the features of the drug and its target. Web applications such as similarity ensemble approach (SEA) and ChemMapper are available to predict drug–target interactions. Various strategies that involve machine learning and deep learning are used to determine the drug target binding affinity such as KronRLS, SimBoost, DeepDTA, WideDTA, DeepAffinity and PADME. Protein And Drug Molecule interaction prediction (PADME) considers the combination of the various features of the drug and target protein as input data and forecasts the interaction strength between the two. Target Fishing (TF) cross-docking approach is performed to identify novel protein-drug interactions. This may be useful for explaining various types of side-effects or suggesting new modes of action. For investigating the effects and mechanisms of action of natural products in treatment of atherosclerosis, sepsis, or migraine, TF has been performed to identify the candidate targets and target compound interactions. The bioactivity of a drug also includes ADME data to get the metabolism idea numerous AI-based tools, like XenoSite, FAME, and SMARTCyp are involved in determining the sites of metabolism of the drug. In addition various softwares such as CypRules, MetaSite, MetaPred, SMARTCyp, and WhichCyp can be used to identify specific isoforms of CYP450 that mediate a particular drug metabolism. In some study it was found that the clearance pathway of 141 approved drugs was done by support vector machines (SVM)-based predictors with high accuracy.
7. AI in Clinical Trial Design
AI can assist in choosing a disease-specific population for recruitment in Phase II and III of clinical trials by using patient-specific genome – exposome profile analysis, that can facilitate in early and fast prediction of the available drug targets in the patients which are selected for the trials. Through predictive machine learning and other reasoning techniques, it helps in the early prediction of lead molecules that would pass clinical trials with consideration of the selected patients. Drop out of patients from clinical trials accounts for the failure of 30 per cent of the clinical trials, creating additional recruiting requirements for the completion of the trial, leading to wastage of time and money. This problem can be overcome by close monitoring of the patients and helping them to follow the desired protocol of the clinical trial. For example mobile software is developed by AiCure that monitors regular medication intake by patients with schizophrenia in a Phase II trial which increased the adherence rate of patients by 25 percent ensuring successful completion of the clinical trial.
AI IN MANUFACTURING OF PHARMACEUTICAL DRUGS
A Manufacturing Execution System (MES) is a control system that is designed to manage, monitor, and track the various manufacturing information in real time by receiving minute by minute data from various sources which include robots, employees, and machine monitors. MES facilitates compliance with regulatory guidelines along with ensuring that drug makers get high-quality products in their manufacturing processes. The benefits of using MES include compliance with guaranteed legal regulations, minimised risks, increased transparency, shortened production cycles, optimised resource utilisation, controlled, and monitored production steps, and optimised up to batch release. Various tools such as computational fluid dynamics (CFD), use Reynolds-Averaged Navier-Stokes solvers technology that studies the impact of agitation and stress levels in several equipments (e.g., stirred tanks), exploiting the automation of many pharmaceutical operations. Direct numerical simulations and large eddy simulations are similar systems that involve advanced approaches to solve complicated flow problems in manufacturing.
The new Chemputer platform helps digital automation for the synthesis and manufacturing of molecules, by incorporating various chemical codes and operating by using a scripting language known as Chemical Assembly. This process has been successfully implemented for the synthesis and manufacture of drugs like sildenafil, diphenhydramine hydrochloride, and rufinamide that are having the yield and purity significantly similar to manual synthesis. Meta-classifier and tablet-classifier are AI tools that facilitate to control the quality standard of the final product, indicating a possible error in the manufacturing of the tablet.
AI IN DRUG DELIVERY
Methods involved in drug delivery are focused ultrasound, micro-pump mechanism, and targeted delivery by microrobots. In 1959, Richard Feynman proposed the idea of using microrobots for medical treatment. Micro/nanorobots have great potential in medical treatment as they can be applied in surgical operation, disease diagnosis, targeted drug delivery etc. They can move autonomously, which makes it possible to deliver drugs to the difficult to find areas. They use an internal payload and an external shell that can reach a specific target actively also several driving modes are often used jointly to manufacture micro/nanorobots with multiple functions that can be driven either by exogenous power (magnetic fields, light energy, acoustic fields, electric fields) or endogenous power (chemical reaction or biological reaction). Example of Magnetic Field-driven Micro/Nanorobots- Qiu et al. designed a microrobot model including two parts: Titanium-coated (artificial bacterial flagella) beneath rotating magnetic fields that could achieve precise 3D-navigation in fluids, it consist of the outermost temperature-sensitive liposome part that could release the drug (Calcein) under the temperature regulation; which confirmed its potential application in the targeted drug delivery. While example of Light Energy driven Micro/Nanorobots-Zhan et al. designed an artificial swimmer by integrating two cross-aligned dichroic nanomotors its movement can be navigated by regulating the incident light’s polarization direction this model might be effective for targeted delivery of drugs.
Apart from the function mode of directly driving the movement of micro/nanorobot, light energy also plays a catalytic role in inducing the redox reaction inside the micro/nanorobot that further propels the nanorobot by producing chemical gradients. The endogenous propulsion method involves the use of power that promotes the self-propelling of nanorobots by chemical reaction or biological reaction; this type of micro/nanorobot is often coated with catalysers to get the continuous chemical energy from the environment.
Besides use of microrobots for drug delivery, artificial neural netwoks (ANN) can also be used in optimising and modelling the complex relationships between the formulation parameters and their effects on the quality of final product. ANNs also provide an idea about release pattern of drug product as shown in the study conducted by Baghaei B. et al. about polylactic-co-glycolic acid (PLGA) Nanoparticles. ANNs help to reduce the prediction error from 28.0 to 2.93 percent and from 19.4 to 2.99 percent for particle size and leads to initial burst release of PLGA Nanoparticles. ANN models can be used to release doxorubicin from Pluronic P-105 micelles.The model showed that more effective drug delivery can occur even at lower frequency, low copolymer concentrations and enhanced power density while thermal effects do not play any important role. Example- ANN is used to make stable emulsion. In a study conducted by Kumar et al. ANNs have been used to formulate stable o/w emulsion and optimising the concentration of a fatty alcohol. For emulsion formulation, input data includes the variables of lauryl alcohol concentrations, time. While, zeta potential, particle size, viscosity, conductance are considered as outputs. To study the formulation of stable emulsion, ANN models have been developed containing isoniazid and rifampicin for oral delivery; several data from pseudo ternary phase triangles containing the mixture of surfactants and oil component were applied for training, testing, and validation of the ANN model. Using the radial basis function network architecture, obtained microemulsion formulation with improved stability was capable of targeted delivery of both anti-tuberculous drugs.
Thus, Artificial Intelligence is the simulation of human intelligence processes by machines, the roots of which were traced back in 1942 and currently it is being used in various areas. Further, the appropriate knowledge and effective implementation of AI is very essential to reap the benefits in the pharmaceutical manufacturing and healthcare sector so as to circumvent different problems and to increase the patient compliance in the treatments of variouss diseases.
1. Nadimpalli M. Artificial intelligence risks and benefits. International Journal of Innovative Research in Science, Engineering and Technology. 2017 Jun;6(6).
2. Sarmah SS. Concept of Artificial Intelligence, its Impact and Emerging Trends. 2019.
3. Wang W, Siau K. Artificial intelligence, machine learning, automation, robotics, future of work and future of humanity: a review and research agenda. Journal of Database Management (JDM). 2019 Jan 1;30(1):61-79.
4. Hassanzadeh P, Atyabi F, Dinarvand R. The significance of artificial intelligence in drug delivery system design. Advanced drug delivery reviews. 2019 Nov 1;151:169-90.
5. Haenlein M, Kaplan A. Guest editorial to the special issue, A brief history of AI: On the past, present, and future of artificial intelligence. California Management Review. 2019;61(4):5-14.
6. Vaishya R, Javaid M, Khan IH, Haleem A. Artificial Intelligence (AI) applications for COVID-19 pandemic. Diabetes & Metabolic Syndrome: Clinical Research & Reviews. 2020 Jul 1;14(4):337-339.
7. Shabbir J, Anwer T. Artificial intelligence and its role in near future. arXiv preprint arXiv:1804.01396. 2018 Apr 1.
8. Vyas M, Thakur S, Riyaz B, Bansal KK, Tomar B, Mishra V. Artificial intelligence:The beginning of a new era in pharmacy profession. Asian J Pharm. 2018 Apr 1;12(2):72-6.
9. Hu M, Ge X, Chen X, Mao W, Qian X, Yuan WE. Micro/Nanorobot: A Promising Targeted Drug Delivery System. Pharmaceutics. 2020 Jul;12(7):665.
10. Egorov E, Pieters C, Korach-Rechtman H, Shklover J, Schroeder A. Robotics, microfluidics, nanotechnology and AI in the synthesis and evaluation of liposomes and polymeric drug delivery systems. Drug Delivery and Translational Research. 2021 Feb 13:1-8
11. Artificial Intelligence.What is Artificial Intelligence? How Does AI Work? https://builtin.com/artificial-intelligence (Access on 17th March 2021).
12. All India Radio launches new series on Artificial Intelligence (AI) https://content.techgig.com/all-india-radio-launches-new-series-on-artificial-intelligence-ai/articleshow/81550453.cms (Access on 17th march 2021).
13. How can Machine learning accelerate the pace of drug discovery?https://www.analyticsinsight.net/how-can-machine-learning-accelerate-the-pace-of-drug-discovery/ (Access on 18th march 2021)
14. Applications of Artificial Intelligence in Pharma Industry https://usmsystems.com/artificial-intelligence-in-pharma/ (Access on 24 march 2021)
15. A Guide to AI in the Pharmaceutical Industry https://stefanini.com/en/trends/news/a-guide-to-ai-in-the-pharmaceutical-industry(Access on 24 march 2021)
16. Paul D, Sanap G, Shenoy S, Kalyane D, Kalia K, Tekade RK. Artificial intelligence in drug discovery and development. Drug Discov Today. 2021 Jan;26(1):80-93.
17. Artificial Intelligence In The Workplace: How AI Is Transforming Your Employee Experience.https://www.forbes.com/sites/bernardmarr/2019/05/29/artificial-intelligence-in-the-workplace-how-ai-is-transforming-your-employee-experience/?sh=2ab2322753ce(Access on 25 march 2021)