Abstract: Chronic diseases are now the leading cause of death worldwide with over 40 million premature deaths. Statistics tell us, for example, that diabetes has more than doubled its prevalence in twenty years, also in relation to the progressive aging of the population. It is estimated that the complex of chronic diseases, also called non-communicable, could cost the economy almost fifty trillion dollars in 2030. 
The demand for health in the Western world has changed compared to the past; today we talk more about diagnosis and treatment or risk assessment, especially of prevention, aging success and statistics applied to the individual. In other words, the patient feels the need for a global takeover by the health system and wants more and more answers on an individual basis and not as a member of a group. The goal to be achieved is not so much to lengthen the life span, but to increase the life span free from age-dependent diseases.There is more and more talk of big data derived from the explosion of the so-called "omics" that are dramatically changing the reference scenario for medicine, allowing a systemic and holistic view of the subject, with the ambition to transfer therapeutic decisions and predictions of outcome at the individual level, starting from big data. 
Emblematic from this point of view is the program launched in the United States by the National Institute of Health to build a huge database that will collect clinical data laboratory and imaging data of one million U.S. citizens who will be followed for 20 years with the aim of developing predictive models of major chronic diseases, from cancer to diabetes to cardiovascular diseases. Big data and their inherent complexity are overwhelmingly calling into question the so-called "Machine learning systems", statistical systems based on artificial intelligence that according to the most authoritative experts will be indispensable to improve the accuracy of medical decisions. Thanks to text mining techniques, another branch of artificial intelligence, even documents in natural language will become computable with the great benefit of reducing the downtime spent on the acquisition of a large amount of clinical information to the advantage of a time spent to affect with a better relationship doctor-patient compliance with the indications given for a change in their lifestyle. 
If this is the scenario in clinical medicine, what about drug discovery in R&D? 
Also in this field the scientific progress of the last decades has allowed the increasingly sophisticated development of Artificial Intelligence (AI) algorithms that, enabled through the large amount of data generated (big data), now represents one of the main guidelines for the future of the drug industry. The activation of AI in the chemical pharmaceutical field has now made reality the development of new targets for the discovery of new drugs. A reality no longer hypothesized in the future, but a practice in the major realities of the world pharmaceutical industry. 
Computer-aided drug design techniques have been therefore successfully applied in almost every stage of the drug discovery and development pipeline with the resulting speed up of R&D process of research and reduction of the costs and risks related to preclinical and clinical trials. Owing to the development of machine learning theory and the accumulation of pharmacological data, the artificial intelligence (AI) technology, as a powerful data mining tool, has cut a figure in various fields of the drug design, such as virtual screening, activity scoring, quantitative structure-activity relationship (QSAR) analysis, de novo drug design, and in silico evaluation of absorption, distribution, metabolism, excretion and toxicity (ADME/T) properties. The immediate results of this revolution is the positive impact on the costs that the company must bear throughout the process of research and development. 
The R&D process, in fact, represents one of the longest, most uncertain and expensive moments of drug development. It can take up to 15 years from initial testing to market transfer, with an average cost of around $2.5 billion and a very high risk of failure. Only 10% of the compounds enter phase 1 of clinical development, while the failure rate in phases 2 and 3 is around 50%. 
In this perspective today the most complex areas in the development of the sector is becoming a fertile ground for the possibilities offered by artificial intelligence and robotics applied to drug R&D. Once integrated, machine learning and data mining capabilities could improve the planning operations of the R&D process, as well as the use of Robotics to rapidly test millions of molecules could allow the development of new solutions unimaginable until recently. 
Deep learning opens today the door to a new path: that of designing new virtual molecules in silico to extrapolate what could be a good candidate for a new drug, to be synthesized materially in the laboratory at a later time. In this direction this virtual chemical assistant explores the molecules intuitively, using the chemical knowledge. The problem is to "teach" the algorithm to work on certain number of known molecules to design new molecules, predict their behavior and suggest possible combinations. Unfortunately, an algorithm to produce excellent results must have a lot of data to work on, and the data cost. It is therefore necessary to develop algorithms that are able to work with less and less data. 
One thing is certain: the pharmaceutical industry is investing heavily in this direction. This is demonstrated by the fact that from 2015 to the present day, pharmaceutical companies such as Merck, Janssen Pharmaceutica, Sanofi, Genentech, Takeda, Santen Pharmaceuticals have entered into collaborations with known realities in the field of artificial intelligence for the development of new therapeutic lines in various areas, from the oncology, to that of infectious diseases, malaria in the first place, up to rare diseases. 
This is undoubtedly an interesting avenue, but before we can see if we can be optimistic we need to have concrete success stories in our hands. We must first see the first car without a driver before we can say that we have the new Tesla from the Pharma in our hands. 

References 

1: Ekins S, Puhl AC, Zorn KM, Lane TR, Russo DP, Klein JJ, Hickey AJ, Clark AM. 
Exploiting machine learning for end-to-end drug discovery and development. Nat 
Mater. 2019 May;18(5):435-441. doi: 10.1038/s41563-019-0338-z. Epub 2019 Apr 18. 
Review. PubMed PMID: 31000803. 

2: Vamathevan J, Clark D, Czodrowski P, Dunham I, Ferran E, Lee G, Li B, 
Madabhushi A, Shah P, Spitzer M, Zhao S. Applications of machine learning in drug 
discovery and development. Nat Rev Drug Discov. 2019 Apr 11. doi: 
10.1038/s41573-019-0024-5. [Epub ahead of print] Review. PubMed PMID: 30976107. 

3: Lee CY, Chen YP. Machine learning on adverse drug reactions for 
pharmacovigilance. Drug Discov Today. 2019 Mar 12. pii: S1359-6446(18)30367-2. 
doi: 10.1016/j.drudis.2019.03.003. [Epub ahead of print] Review. PubMed PMID: 
30876845. 

4: Schneider G, Clark DE. Automated De Novo Drug Design: Are We Nearly There Yet? 
Angew Chem Int Ed Engl. 2019 Feb 7. doi: 10.1002/anie.201814681. [Epub ahead of 
print] Review. PubMed PMID: 30730601. 

5: Neves BJ, Braga RC, Melo-Filho CC, Moreira-Filho JT, Muratov EN, Andrade CH. 
QSAR-Based Virtual Screening: Advances and Applications in Drug Discovery. Front 
Pharmacol. 2018 Nov 13;9:1275. doi: 10.3389/fphar.2018.01275. eCollection 2018. 
PubMed PMID: 30524275; PubMed Central PMCID: PMC6262501. 

6: Maltarollo VG, Kronenberger T, Espinoza GZ, Oliveira PR, Honorio KM. Advances 
with support vector machines for novel drug discovery. Expert Opin Drug Discov. 
2019 Jan;14(1):23-33. doi: 10.1080/17460441.2019.1549033. Epub 2018 Nov 29. 
PubMed PMID: 30488731. 

7: Sharma S, Sharma D. Intelligently Applying Artificial Intelligence in 
Chemoinformatics. Curr Top Med Chem. 2018;18(20):1804-1826. doi: 
10.2174/1568026619666181120150938. Review. PubMed PMID: 30465503. 

8: Pérez-Sianes J, Pérez-Sánchez H, Díaz F. Virtual Screening Meets Deep 
Learning. Curr Comput Aided Drug Des. 2019;15(1):6-28. doi: 
10.2174/1573409914666181018141602. Review. PubMed PMID: 30338743. 

9: Dana D, Gadhiya SV, St Surin LG, Li D, Naaz F, Ali Q, Paka L, Yamin MA, 
Narayan M, Goldberg ID, Narayan P. Deep Learning in Drug Discovery and Medicine; 
Scratching the Surface. Molecules. 2018 Sep 18;23(9). pii: E2384. doi: 
10.3390/molecules23092384. Review. PubMed PMID: 30231499; PubMed Central PMCID: 
PMC6225282. 

10: Zhang W, Lin W, Zhang D, Wang S, Shi J, Niu Y. Recent advances in the machine 
learning-based drug-target interaction prediction. Curr Drug Metab. 2018 Aug 20. 
doi: 10.2174/1389200219666180821094047. [Epub ahead of print] PubMed PMID: 
30129407. 

11: Panteleev J, Gao H, Jia L. Recent applications of machine learning in 
medicinal chemistry. Bioorg Med Chem Lett. 2018 Sep 15;28(17):2807-2815. doi: 
10.1016/j.bmcl.2018.06.046. Epub 2018 Jun 28. Review. PubMed PMID: 30122222. 

12: Rifaioglu AS, Atas H, Martin MJ, Cetin-Atalay R, Atalay V, Dogan T. Recent 
applications of deep learning and machine intelligence on in silico drug 
discovery: methods, tools and databases. Brief Bioinform. 2018 Jul 31. doi: 
10.1093/bib/bby061. [Epub ahead of print] PubMed PMID: 30084866. 

13: Zhong F, Xing J, Li X, Liu X, Fu Z, Xiong Z, Lu D, Wu X, Zhao J, Tan X, Li F, 
Luo X, Li Z, Chen K, Zheng M, Jiang H. Artificial intelligence in drug design. 
Sci China Life Sci. 2018 Oct;61(10):1191-1204. doi: 10.1007/s11427-018-9342-2. 
Epub 2018 Jul 18. Review. PubMed PMID: 30054833.