1000/1000
Hot
Most Recent
COVID-2019 pandemic lead to a raised interest on the development of new treatments through Artificial Intelligence (AI). AI is a suitable tool to quickly analyze large amounts of data or to estimate drug repurposing against COVID-2019.
Author(s); Year Reference |
Country(ies) of Origin | Key Repurposed Drug(s) Potentially Relevant against the Treatment of COVID-2019 | Key Used AI Methodologies |
---|---|---|---|
(Ke, Peng, Yeh, et al., 2020) [13] | Taiwan | Bedaquiline Brequinar Celecoxib Clofazimine Conivaptan Gemcitabine Tolcapone Vismodegib |
First, an an AI platform was defined to identify potential old/repurposed drugs with anti-coronavirus activities (or potential anti-coronavirus activity). Second, AI predicted drugs were tested for their activity against a feline coronavirus in vitro. Third, results of assays were introduced in the AI system. A Deep Neural Network algorithm was used to identify the most relevant descriptors, with the generation of different weightings to generate AI prediction models. |
(Morselli, do Valle, Zitnik, et al., 2021) [14] | USA, Turkey, Hungary | Auranofin Azelastine Fluvastatin Methotrexate Vinblastine |
AI based on algorithms, network diffusion, and network proximity, tasking to rank 6,340 drugs, regarding their potential efficacy against SARS-CoV-2. Multimodal technology was required to fuse the prediction of all algorithms, since the predictive algorithm did not offer consistently reliable outcomes. The top-ranked drugs were screened in human cells. |
(Abdulla, Wang, Qian, et al., 2020) [15] * | China, Singapore | Amantadine Azithromycin Chloroquine Omeprazole Sodium (optional) Ribavirin (optional) - These repurposing drugs were evaluated in combination. |
An AI-based platform was used to interrogate 12 drug/dose parameters space. Combination therapies that optimally inhibit A549 lung cell infection by vesicular stomatitis virus within 3 days of project start were identified. This AI project utilized a quadratic relationship between drug/dose inputs and efficacy/safety outputs to successfully identify the drug-dose parameter space. |
(Blasiak, Lim, Seah, et al., 2020) [16] * | China, Singapore, USA | Lopinavir Remdesivir Ritonavir - These repurposing drugs were evaluated in combination. |
A platform (IDentif.AI) that pairs experimental validation with AI and digital drug development was used. Workflow of the project IDentif.AI: 1) clinically relevant concentrations based on and dose–response curves and maximal plasma concentration; 2) in vitro testing of combination therapies; combination therapies were determined through an orthogonal array composite (OACD) design; 3) IDentif.AI analysis: drug–drug interactions and clinically relevant drug-dosage combinations; and 4) biological validation. |
Studies with confirmatory in-vitro and/or clinical data | |||
(Schultz, Vera, Sinclair et al., 2020) [17] | USA | Baricitinib | |
(Stebbing, Krishnan, de Bono et al., 2020) [18] | UK, USA, Italy, Sweden, Singapore | Baricitinib | BenevolentAI (an artificial AI platform) identified baricitinib as a potential COVID-19 drugs. Details/information on BenevolentAI works were limited. |
Repurposing of drugs against COVID-2019 | |||
(Nayarisseri, Khandelwal, Madhavi et al., 2020) [19] | India, Saudi Arabia | Aprepitant Fulvestrant Remdesivir Valrubicin |
A machine learning approach was employed. Particularly, repurposed drugs were selected based on their capacity of targeting the main coronavirus protease (6LU7) using ligand-receptor Docking (optimized potential for liquid simulations algorithms to identify high affinity compounds). Additionally, candidates were subjected to Molecular Dynamic Simulations followed by ADMET (absorption, distribution, metabolism, excretion, and toxicity) studies. |
(Kim, Zhang, Cha et al., 2020) [20] | USA | Emricasan Fosamprenavir Glutamine Glutathione Piperacillin sodium Ruxolitinib Vitamin E |
Two computational approaches were applied. Fist, a high-throughput AI-based binding affinity prediction platform was used to identify FDA approved drugs with potential capacity to block coronaviruses from entering cells by binding to ACE2 (angiotensin-converting enzyme) or TMPRSS2 (Transmembrane Serine Protease 2). Second, the Disease Cancelling Technology (DCT) platform was used to identify FDA approved drugs, which may attenuate the gene expression patterns induced by coronaviruses. |
(Das G., Das, T., Chowdhury et al., 2021) [21] | India | Ascorbyl palmitrate Cinametic acid Guaifenesin Lauric acid Nabumetone Nafcillin Octacosanol Palmidrol Salmeterol |
AI deep learning techniques, in silico methodologies) and pattern recognition techniques were used to screen FDA approved pharmaceuticals and nutraceuticals to target CoV envelope (E) protein. A protein involved in the assembly and release of the virus inside the host. Multiple opensource drug databases were considered, such as ChEMBL v.26, Enamine Bio reference Compounds (https://www.enaminestore.com/products/bioreference-compounds, accessed on 17 September 2021) and Chemoinformatic tools and database (https://chemoinfo.ipmc.cnrs.fr/TMP/tmp.32396/e-Drug3D_1930_v3.sdf, accessed on 17 September 2021). |
(Rajput, Thakur, Mukhopadhyay et al., 2021) [22] | India | Alatrofloxacin Metergoline Rescinnamine Rescinnamine Telotristat ethyl Verteporfin |
Robust computational methods using machine learning techniques, such as Support Vector Machine, Random Forest, k-Nearest Neighbour, Artificial Neural Network, and Deep Learning were developed by the authors to predict the repurposed drugs. |
(Li, Yao, Cheng et al., 2021) [23] | China, USA | Baricitinib Bivalirudin Fostamatinib Lusutrombopag Simvastatin |
1) Public genetic screening data were successively interrogated to identify human-specific host dependency genes, i.e., indispensable genes for effective viral infections. 2) Extensive drug-target interactions were interrogated through diverse methodologies, such as database retrieval, literature mining and de novo prediction using AI-based algorithms. |
(McCoy, Gudapati, He, Horlander, 2021) [24] | USA | Amprenavir Albuterol Artemisinin Chloroquine Ciprofloxacin Cyclosporine Fluoroquinolones Hydroxymethylglutaryl-CoA reductase inhibitors Methotrexate Quinolone antibacterial agents Suramin Zidovudine |
A link prediction model was developed (an AI text mining model). The biomedical knowledge graph, SemNe was used to predict missing links in biomedical literature, regarding drug repurposing. TransE, CompleX, and RotatE based methods were used to visualize knowledge graph embeddings and link prediction results using in a web application. |
(Chakravarty, Antontsev, Khotimchenko et al., 2021) [25] | USA | Captopril Lisinopril Spirapril |
The plataform BIOiSIM (an AI-integrated mechanistic modeling platform) was used to simulate systemic therapy of Calcium Channel Blockers (CCBs) and ACE compounds in tissues related to the COVID-19 pathogenesis, namely the disposition and site-of-action penetration (in silico modeling). BIOiSIM is a dynamic, biology-driven platform that provides a scalable computational prediction of in vivo pharmacokinetic-pharmacodynamic (PK-PD) phenomena. |
(Kadioglu, Saeed & Efferth, 2021) [26] | Germany | Conivaptan Dihydroergotamine Eltrombopag Ergotamine Eribulin Idarubicin Ivermectin Ledipasvir Lifitegrast Lumacaftor Nilotinib Nystatin Paritaprevir Ponatinib Regorafenib Rifapentine Simeprevir Teniposide Trabectedin Trypan blue Velpatasvir Venetoclax |
Diverse combined in silico methods (virtual drug screening, molecular docking, and supervised machine learning algorithms) were used in a workflow to identify repurposed drug against COVID-19. |
Repurposing of drugs against COVID-2019: association of drugs | |||
(Artigas, Coma, Matos-Filipe et al., 2020) [27] | Spain | Pirfenidone plus melatonin | The mechanism of action of pirfenidone and melatonin was simulated by using the previously described Therapeutic Performance Mapping System (TPMS) technology (an AI-based approach). GUILDify v2.0 web server was used to confirm the effect of pirfenidone and melatonin against SARS-CoV-2 infection. This web server is able to calculate the neighbourhoods of the human biological network related to the host-key points (e.g., for SARS-CoV infection) and simultaneously affected by specific drugs. |
Note: References [15][16] are also related to the combination of repurposing medicines. | |||
Repurposing of drugs: alternative therapies (Section 3.2.3.2) | |||
(Wang, Li; Song, et al., 2021) [28] | China, Australia | GCT-CJ Hanshiyufen fang (HSYF-F) Huashi Baidu Formula (HSBD-F) Qingfei Paidu Decoction (QFPD-T) Shenfu zhusheye (SF-ZSY) Zhongqifangzi (PMSP) Recommended as supplementary treatment against COVID-2019. |
An ontology-based side-effect prediction framework (OSPF) was developed based on a previous work and Artificial Neural Network (ANN)-based deep learning. The Traditional Chinese Medicine prescriptions for the treatment of COVID-19 (officially recommended in China) were evaluated. |
(Li, Yao, Cheng et al., 2021) [23] | China, USA | Atropine (Lycii Cortex, Hyoscyami Semen) Dehydroeffusal (Junci Medulla) Lysergol (Pharbitidis Semen) Solanocapsine (Solanum Nigrum) Solanocapsine (Solanum Nigrum) Vitexifolin C (Viticis Fructus) |
1) Public genetic screening data were successively interrogated to identify human-specific host dependency genes, i.e., indispensable genes for effective viral infections. 2) Extensive drug-target interactions were interrogated through diverse methodologies, such as database retrieval, literature mining and de novo prediction using AI-based algorithms. |
(Kadioglu, Saeed & Efferth, 2021) [26] | Germany | Amyrin Baicalin Crinine Euphol Forsythiaside Friedelin Hoslunddiol IlexsaponinB1 IlexsaponinB2 IlexsaponinB3 Loniflavone Procyanidin Punicalagin Quercetin Quercetin-3-o-rutinoside Rutin Strictinin TirucallinA TingeninB Wogonoside ZINC252515584 ZINC27215482 ZINC15675938 |
Diverse combined in silico methods (virtual drug screening, molecular docking, and supervised machine learning algorithms) were used in a workflow to identify repurposed drug against COVID-19. |
(Acharya, Agarwal, Baker et al., 2020) [29] | USA, Italy | Hypericin Quercetin |
An enhanced sampling molecular dynamics (MD) and ensemble docking was used supercomputer-driven pipeline for in silico drug discovery. |