{
  "id": "anticonvulsant",
  "title": "Anticonvulsant",
  "lang": "en",
  "source_lang": "en",
  "summary": "Anticonvulsants are medicines that suppress epileptic seizures by modifying neuronal excitability, with additional uses in pain, mood disorders, and migraine prevention.",
  "category": "health",
  "subcategory": "drugs",
  "type": "concept",
  "aliases": [
    "Antiseizure medication",
    "Antiseizure drug",
    "Antiepileptic drug",
    "Anticonvulsive drug"
  ],
  "generated_at": "2026-11-05",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/anticonvulsant/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/anticonvulsant/",
      "json": "https://aiwiki.page/zh/anticonvulsant.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "An anticonvulsant is a medicine used to prevent or terminate epileptic seizures, episodes of abnormal, excessive or synchronized electrical activity in the brain. Anticonvulsants are central to the treatment of epilepsy, but they also have roles in acute neurological emergencies and, for selected drugs, in neuropathic pain, bipolar disorder, and migraine prevention. They comprise a pharmacologically diverse group rather than a single chemical family.\n\nThe term antiseizure medication increasingly replaces anticonvulsant and antiepileptic drug. “Anticonvulsant” can misleadingly imply that treatment concerns only seizures with visible convulsions, whereas many seizures produce altered awareness, sensory experiences, or brief interruptions of activity without major movements. “Antiepileptic” can imply modification of the underlying disease, although most established medicines suppress seizures without demonstrably eliminating the enduring predisposition to them.\n\nThe clinical effects of these medicines depend on their molecular targets, the seizure type and epilepsy syndrome, and the characteristics of the person receiving treatment. Their study therefore spans pharmacology, clinical neurology, developmental medicine, and medication safety. Important distinctions include ongoing seizure prevention versus emergency seizure termination, broad-spectrum versus more selective activity, and symptomatic seizure suppression versus treatment directed at an underlying cause."
    },
    {
      "heading": "Historical development and changing terminology",
      "text": "The modern history of anticonvulsant treatment is commonly traced to Charles Locock’s report on potassium bromide in 1857. Bromides became widely used before the physiological basis of epilepsy was understood. They could reduce seizures, but their long persistence in the body and dose-related toxicity produced sedation, cognitive impairment, skin eruptions, and the chronic intoxication syndrome known as bromism. Their use established that chemical treatment could influence seizure susceptibility, even when explanations for the effect were initially incorrect.\n\nPhenobarbital entered epilepsy treatment in 1912 following Alfred Hauptmann’s observations. It was more practicable than bromide treatment and remains in use, particularly where cost and availability strongly influence access. Its sedative properties, however, illustrated a recurring problem: suppression of seizures had to be distinguished from nonspecific suppression of brain function. A medicine could prevent convulsions yet impose substantial limitations through drowsiness or impaired coordination.\n\nA major shift occurred in 1938, when H. Houston Merritt and Tracy Putnam reported the anticonvulsant activity of phenytoin. Their experimental screening used electrically induced seizures and helped establish systematic animal testing as a route to drug discovery. Phenytoin demonstrated that strong protection against certain seizures did not require the degree of sedation associated with barbiturates. Subsequent development produced drugs with different seizure spectra, including ethosuximide for absence seizures, carbamazepine for focal seizures, and valproate as a broad-spectrum medicine. Valproate’s anticonvulsant activity was recognized in France in 1962 after it had been used as a solvent in experimental testing.\n\nFrom the late twentieth century onward, development increasingly incorporated defined molecular targets and more differentiated clinical testing. Vigabatrin was designed around inhibition of the enzyme that degrades the inhibitory transmitter GABA. Levetiracetam was developed before its principal binding target, synaptic vesicle protein 2A, was identified. Later drugs included selective modulators of ion channels, glutamate receptors, and synaptic proteins. These developments expanded treatment options but did not abolish drug resistance.\n\nHistorical labels such as “first-generation,” “second-generation,” and “third-generation” describe approximate periods of introduction rather than a consistent hierarchy of efficacy or safety. Newer medicines often have fewer metabolic interactions or more convenient pharmacokinetics, but an older medicine can remain particularly effective for a specific seizure type. Contemporary terminology consequently emphasizes what a drug demonstrably does—suppress seizures—rather than assuming that its date of introduction predicts its clinical value."
    },
    {
      "heading": "Molecular mechanisms and neuronal networks",
      "text": "Seizures emerge from disturbances in the excitability and synchronization of neuronal networks. Anticonvulsants influence this activity by altering ion channels, neurotransmitter release, receptor function, or inhibitory signaling. Their effects cannot be reduced to a universal “calming” action: different seizure networks respond differently, and some medicines effective against one seizure type can aggravate another.\n\nSeveral important drugs act on voltage-gated sodium channels, which generate the rapid electrical changes underlying neuronal action potentials. Phenytoin, carbamazepine, and lamotrigine preferentially limit sustained, repetitive firing by interacting with channels in particular functional states. This use-dependent action makes their effects more pronounced during rapid neuronal activity than during ordinary signaling. Lacosamide is generally described as enhancing slow inactivation of sodium channels. These distinctions matter because drugs grouped under the same broad mechanism need not have identical physiological or clinical effects.\n\nOther medicines strengthen inhibition mediated by gamma-aminobutyric acid (GABA). Benzodiazepines enhance the effect of GABA at GABA type A receptors, increasing inhibitory chloride-channel activity in susceptible neuronal systems. Phenobarbital acts at a different site on the same receptor complex. Tiagabine inhibits GABA uptake, whereas vigabatrin irreversibly inhibits GABA transaminase, reducing GABA degradation. These mechanisms increase inhibitory influence in different ways and have different consequences for tolerability, withdrawal, and seizure specificity. Inhibitory signaling also depends on neuronal chloride gradients, so GABA’s physiological effects vary with developmental and pathological context.\n\nCalcium channels provide additional targets. Ethosuximide reduces T-type calcium currents involved in the thalamocortical rhythms associated with absence seizures. Gabapentin and pregabalin bind the alpha-2-delta auxiliary subunit of voltage-gated calcium channels and reduce activity-dependent transmitter release; despite their names and structural relationship to GABA, they do not act primarily as GABA-receptor agonists. Levetiracetam and brivaracetam bind synaptic vesicle protein 2A, modifying processes involved in neurotransmitter release. Perampanel antagonizes AMPA-type receptors for glutamate, the principal excitatory neurotransmitter in the central nervous system.\n\nSome medicines have several relevant actions. Valproate affects sodium-channel function and GABA-related processes, among other effects. Topiramate combines actions on sodium channels, GABA-mediated inhibition, glutamate receptors, and carbonic anhydrase. Such classifications are useful but incomplete: the relative contribution of each action at clinical concentrations may remain uncertain. Moreover, the same molecular target can participate in distinct circuits. A drug’s mechanism therefore helps explain its activity but does not, by itself, establish which epilepsy syndromes it will treat successfully."
    },
    {
      "heading": "Drug groups and seizure-specific activity",
      "text": "Anticonvulsants can be classified by chemical structure, molecular target, clinical seizure spectrum, or route of administration. None of these systems is sufficient alone. Structurally related medicines can differ in interactions or tolerability, while chemically unrelated medicines can share a target. The practical distinction between broad-spectrum and narrower-spectrum activity refers mainly to the seizure types against which a drug is effective, not to its overall potency.\n\nFocal seizures begin within networks limited to one cerebral hemisphere, although they may spread into bilateral tonic–clonic seizures. Medicines with established activity against focal seizures include carbamazepine, oxcarbazepine, lamotrigine, levetiracetam, lacosamide, and several others. Their comparative usefulness depends on seizure control, adverse effects, interactions, and the clinical population studied. A favorable result in one setting—for example, adjunctive treatment in adults with difficult-to-treat focal epilepsy—does not automatically establish superiority as initial treatment or in children.\n\nGeneralized epilepsies involve networks that rapidly engage both sides of the brain. Valproate has activity across generalized tonic–clonic, absence, and myoclonic seizures, but reproductive and other safety concerns substantially constrain its use. Levetiracetam, lamotrigine, topiramate, and other agents have roles in particular generalized seizure types. “Broad-spectrum” does not mean equally effective against every generalized syndrome: lamotrigine, for example, can aggravate myoclonus in some individuals.\n\nAbsence epilepsy illustrates why precise classification matters. Ethosuximide is specifically effective against typical absence seizures but does not provide the broader protection needed for coexisting generalized tonic–clonic seizures. In the Childhood Absence Epilepsy trial published in 2010, ethosuximide and valproic acid were more effective than lamotrigine for initial control, while ethosuximide produced fewer attentional adverse effects than valproic acid. Conversely, carbamazepine and several other drugs primarily used for focal epilepsy can worsen absence or myoclonic seizures. Similar-looking episodes therefore cannot be assumed to require the same medication.\n\nDevelopmental and epileptic encephalopathies require additional syndrome-specific distinctions. In Dravet syndrome, commonly associated with pathogenic variants in SCN1A, several sodium-channel-blocking medicines can aggravate seizures. Stiripentol, pharmaceutical cannabidiol, and fenfluramine have established roles in this syndrome in particular jurisdictions and age groups. Lennox–Gastaut syndrome has its own evidence base for agents such as rufinamide, clobazam, cannabidiol, and others. Infantile epileptic spasms are another distinct setting: hormonal treatments and vigabatrin occupy roles that cannot be inferred from ordinary focal-seizure treatment. Regulatory authorization, formulation, and approved age ranges vary between countries, so membership in the anticonvulsant class does not imply interchangeable indications."
    },
    {
      "heading": "Long-term treatment and assessment of response",
      "text": "Long-term antiseizure treatment aims principally at sustained seizure freedom without unacceptable adverse effects. The decision to begin maintenance treatment is distinct from the treatment of a seizure already in progress. A single seizure associated with a reversible metabolic disturbance does not necessarily imply an enduring epileptic disorder, and seizure suppression does not substitute for establishing the diagnosis or identifying an underlying cause.\n\nClinical classification draws on the event history, witness descriptions, electroencephalography, and, where relevant, magnetic resonance imaging. These investigations help distinguish focal from generalized epilepsy and identify structural or syndromic features. They also help separate epileptic seizures from conditions that anticonvulsants do not directly treat, including many forms of syncope and functional seizures. An apparent lack of drug response may reflect incorrect classification, missed doses, unsuitable treatment, or events that are not epileptic.\n\nInitial treatment commonly uses one medicine, because monotherapy makes effects easier to attribute and generally reduces interaction burden. When seizures persist, clinical studies distinguish replacement monotherapy from add-on, or adjunctive, treatment. Combining medicines can improve seizure control, but combining similar adverse-effect profiles can also increase dizziness, double vision, sedation, or impaired coordination. Mechanistically different drugs are not automatically complementary, and plausible combinations require clinical evidence.\n\nThe International League Against Epilepsy’s 2010 consensus definition describes drug-resistant epilepsy as failure of adequate trials of two tolerated, appropriately chosen and used medication schedules, alone or in combination, to achieve sustained seizure freedom. This definition is important because persistent seizures after two suitable regimens represent a different clinical problem from failure of an unsuitable or inadequately used medicine. Alternatives may include epilepsy surgery, neuromodulation, or a ketogenic diet in selected settings. Resistance to medication does not necessarily mean that no further treatment can work.\n\nResponse assessment extends beyond counting seizures. Seizure severity, injuries, cognition, behavior, school or occupational functioning, and health-related quality of life all affect treatment outcomes. Longitudinal studies show that some people achieve prolonged remission while others have relapsing or persistently active epilepsy. Discontinuation studies consequently examine recurrence risk rather than assuming that seizure freedom proves cure. Abrupt withdrawal can precipitate seizures, particularly with benzodiazepines and barbiturates, and published withdrawal studies generally evaluate planned reductions. Factors associated with recurrence include epilepsy syndrome, prior seizure history, EEG findings, and structural abnormalities; no single feature provides an infallible prediction."
    },
    {
      "heading": "Emergency seizure termination and status epilepticus",
      "text": "Emergency anticonvulsant treatment differs from maintenance therapy in its time scale, routes of delivery, and outcome measures. A medicine that prevents seizures when taken regularly may be unsuitable for terminating an ongoing seizure because it acts too slowly or lacks an appropriate formulation. Conversely, a rapidly acting rescue medicine may be poorly suited to daily maintenance because of sedation, tolerance, or a short duration of action.\n\nStatus epilepticus is a condition in which seizure activity is abnormally prolonged or repeated without adequate recovery. For generalized convulsive status epilepticus, the International League Against Epilepsy’s 2015 operational framework identifies 5 minutes as the point beyond which a seizure should be regarded as abnormally prolonged, and 30 minutes as a time associated with increasing risk of lasting consequences. These time points are specific to this seizure form and are not interchangeable with thresholds for all nonconvulsive states.\n\nBenzodiazepines are established initial medicines for convulsive status epilepticus. Intravenous lorazepam and diazepam, intramuscular midazolam, and buccal or intranasal preparations have different practical roles depending on setting and access. The RAMPART trial, published in 2012, showed that intramuscular midazolam delivered by emergency personnel was at least as effective as intravenous lorazepam under the trial conditions. Its operational advantage was that administration did not require intravenous access. This illustrates how route and speed of delivery can matter as much as intrinsic drug activity.\n\nFor benzodiazepine-refractory convulsive status epilepticus, intravenous levetiracetam, fosphenytoin, and valproate are widely studied options. The Established Status Epilepticus Treatment Trial, reported in 2019, found broadly similar efficacy among these three treatments, with successful seizure cessation and improved responsiveness in approximately half of participants. The result did not establish that they are interchangeable in every patient: cardiac conduction abnormalities, liver disease, pregnancy-related considerations, and other factors affect their suitability.\n\nPersistent seizures despite initial and subsequent treatment constitute refractory status epilepticus and may require intensive care, continuous anesthetic infusions, mechanical ventilation, and continuous EEG monitoring. Visible convulsions can cease while electrographic seizure activity continues, making EEG particularly important. At the same time, the underlying cause—such as infection, toxic exposure, stroke, or an acute metabolic disturbance—requires separate evaluation and treatment. Seizure clusters, consisting of discrete seizures occurring unusually close together, are another setting for rescue formulations, but they are not synonymous with status epilepticus."
    },
    {
      "heading": "Pharmacokinetics, interactions, and concentration monitoring",
      "text": "The clinical behavior of an anticonvulsant depends not only on its target but also on its pharmacokinetics: absorption, distribution, metabolism, and elimination. Oral preparations include immediate-release and extended-release formulations, liquids, and dispersible products. Some drugs also have intravenous formulations. Differences in release characteristics or absorption mean that products containing the same active substance are not always used in exactly the same way.\n\nOlder medicines provide prominent examples of metabolic interactions. Carbamazepine, phenytoin, phenobarbital, and primidone induce drug-metabolizing enzymes and can lower exposure to numerous other medicines. Carbamazepine also induces its own metabolism, so its clearance changes during the early weeks of treatment. Valproate inhibits several metabolic pathways and can markedly increase lamotrigine exposure. Introducing or withdrawing an interacting drug can therefore change concentrations even when the dose of the affected medicine is unchanged.\n\nNot all newer drugs are free of interactions, but some depend less on hepatic enzyme metabolism. Levetiracetam is eliminated substantially through the kidneys and has relatively few clinically important metabolic interactions. Gabapentin and pregabalin are also predominantly renally eliminated. Their exposure rises when renal clearance falls. In contrast, the handling of other agents depends more heavily on liver function, active metabolites, or specific enzyme pathways. Enzyme-inducing anticonvulsants can reduce exposure to hormonal contraceptives, anticoagulants, immunosuppressants, and some antiviral or anticancer medicines. Estrogen-containing contraceptives can substantially increase lamotrigine clearance, making this a bidirectional area of interaction rather than a simple class effect.\n\nPhenytoin illustrates nonlinear pharmacokinetics: as metabolic capacity becomes saturated, a relatively small increase in dose can produce a disproportionate rise in blood concentration. Protein binding adds another complication. Phenytoin and valproate are highly bound to plasma proteins, but the unbound fraction is the portion available for distribution and pharmacological action. Low albumin concentrations, severe illness, renal dysfunction, or competing protein-bound substances can change the relationship between total and unbound concentrations. An apparently conventional total concentration can therefore coexist with an unusually high active fraction.\n\nTherapeutic drug monitoring measures concentrations to answer defined clinical questions, such as suspected toxicity, altered clearance, interactions, or changes during pregnancy. A laboratory reference range is a population guide, not an absolute boundary between efficacy and failure. Some individuals remain seizure-free below a conventional range, while others experience adverse effects within it. The meaning of a result depends on sampling time, formulation, adherence, protein binding, and the person’s previous concentration–response relationship. Routine concentration testing consequently has different value for different drugs and clinical circumstances."
    },
    {
      "heading": "Adverse effects and pharmacogenetic susceptibility",
      "text": "Anticonvulsant adverse effects range from predictable dose-related symptoms to rare, severe reactions. Common neurological effects include sleepiness, dizziness, double vision, imbalance, and slowed information processing. Their frequency often changes with dose, speed of introduction, and concurrent medication. These effects are clinically important because epilepsy itself, underlying brain disorders, and other medicines can produce similar symptoms, complicating attribution.\n\nIndividual drugs have characteristic adverse-effect profiles. Topiramate can cause word-finding difficulty, cognitive slowing, metabolic acidosis, and kidney stones. Carbamazepine and oxcarbazepine can cause hyponatremia. Valproate is associated with weight gain, tremor, thrombocytopenia, and, less commonly, serious liver injury or pancreatitis. Long-term phenytoin exposure can cause gingival overgrowth and other chronic complications. Vigabatrin can produce irreversible visual-field loss. Felbamate’s use is constrained by rare aplastic anemia and severe hepatic injury. These differences preclude treating “anticonvulsant toxicity” as a single uniform syndrome.\n\nSevere cutaneous reactions are especially important in pharmacogenomics. The HLA-B15:02 allele is strongly associated with carbamazepine-related Stevens–Johnson syndrome and toxic epidermal necrolysis in susceptible populations, particularly among some Asian ancestry groups. HLA-A31:01 is associated with a broader range of carbamazepine hypersensitivity reactions. The clinical meaning of a genetic test depends on the particular drug, allele, outcome, and ancestry-related allele frequency. A negative result for one allele does not exclude all serious drug reactions, and broad racial categories are imperfect substitutes for genetic information.\n\nBehavioral and psychiatric effects also differ among medicines. Levetiracetam can cause irritability and other behavioral symptoms in some individuals; perampanel carries warnings about serious psychiatric and behavioral reactions. Other drugs, such as lamotrigine and valproate, have established psychiatric indications, but this does not mean that their behavioral effects are uniformly beneficial. In 2008, the US Food and Drug Administration required warnings concerning suicidal thoughts or behavior for antiepileptic drugs following a pooled analysis of clinical trials. Interpretation of class-wide risk is complicated by differences between drugs, indications, and underlying disorders.\n\nLong-term safety includes bone and metabolic health. Several enzyme-inducing agents are associated with altered vitamin D metabolism and increased risk of reduced bone density; fracture risk also reflects falls, seizures, age, and other factors. Rare adverse reactions may not become evident in preauthorization trials, making pharmacovigilance, registries, and large observational studies essential. Such evidence complements rather than replaces randomized trials, because uncommon harms and effects emerging after years of exposure require different methods of detection."
    },
    {
      "heading": "Pregnancy, development, and aging",
      "text": "Anticonvulsant treatment during pregnancy involves risks arising from both medication exposure and uncontrolled maternal seizures. These risks differ markedly between drugs and are influenced by dose, combination treatment, epilepsy severity, and other factors. Pregnancy registries have been particularly important because randomized comparisons involving potentially harmful fetal exposures are generally not feasible.\n\nValproate has the clearest association with major congenital malformations and adverse neurodevelopmental outcomes among widely used antiseizure medicines. Registries commonly report major malformation frequencies around 10% after valproate monotherapy, although estimates vary with dose and study population. Reported malformations include neural-tube defects, cardiac defects, and other structural abnormalities. Prenatal exposure is also associated with lower cognitive performance and increased rates of developmental disorders. These findings have led to substantial regulatory restrictions and pregnancy-prevention requirements in several jurisdictions. Regulatory language differs across countries and has changed over time.\n\nLamotrigine and levetiracetam have shown lower major-malformation rates than valproate in large pregnancy registries, but “lower risk” does not mean absence of risk. Polytherapy cannot be evaluated simply by counting drugs: the particular agents and their doses matter, with valproate-containing combinations carrying particular concern. The role of folate is also more specific than a general guarantee of protection. Folate is important for neural-tube development, but supplementation has not been shown to remove the full range of medication-associated fetal risks.\n\nPregnancy changes drug handling. Increased renal clearance can lower levetiracetam concentrations, while increased glucuronidation can substantially lower lamotrigine concentrations. After delivery, these changes reverse, potentially increasing exposure if the pregnancy-associated regimen remains unchanged. Drug passage into breast milk varies with protein binding, lipid solubility, and elimination; infant exposure also depends on feeding patterns and the infant’s ability to clear the medicine. Placental transfer and transfer through milk are distinct processes and do not yield identical risk profiles.\n\nAt the beginning and end of life, physiology introduces different considerations. Neonatal seizures often arise from acute brain injury or metabolic disease and may be detectable only by EEG; their treatment evidence differs from that for childhood or adult epilepsy. Phenobarbital remains an established initial medicine in many neonatal protocols, while selected genetic epilepsies respond differently to sodium-channel-active drugs. In older adults, reduced renal function, multiple concurrent medicines, and heightened susceptibility to sedation or imbalance alter treatment tolerability. Epilepsy after stroke is also common in this age group. Across the life course, age alone is less informative than the combination of developmental stage, organ function, underlying syndrome, and interacting treatments."
    },
    {
      "heading": "Other indications, comparative evidence, and new directions",
      "text": "Several anticonvulsants are used outside epilepsy because neuronal excitability and synaptic signaling also contribute to pain, mood regulation, and headache disorders. These uses are drug-specific. Demonstrated benefit for one member of the group does not establish a class effect, and a plausible anticonvulsant mechanism does not establish efficacy for an unrelated symptom.\n\nGabapentin and pregabalin have established roles in selected forms of neuropathic pain, including painful diabetic neuropathy and postherpetic neuralgia. Carbamazepine is an established treatment for trigeminal neuralgia. These conditions differ biologically and clinically from ordinary nociceptive pain. Evidence for anticonvulsants in nonspecific musculoskeletal pain or low-back pain is substantially less favorable, and seizure efficacy provides no basis for assuming general analgesic activity. Gabapentinoids also have recognized misuse potential and can contribute to respiratory depression, particularly with opioids or other central nervous system depressants.\n\nIn bipolar disorder, valproate and carbamazepine have roles in treating mania, while lamotrigine has a role in maintenance treatment, particularly in reducing depressive recurrences. Lamotrigine is not an established treatment for acute mania. Topiramate and valproate also have established preventive uses in migraine, but prevention is different from stopping an individual migraine attack. The benefits and reproductive restrictions relevant to these drugs remain important when the indication is not epilepsy.\n\nComparative research measures more than short-term seizure reduction. Adjunctive trials frequently report the proportion of participants with at least a 50% reduction in seizure frequency, a useful regulatory outcome that is not equivalent to seizure freedom. Pragmatic studies such as SANAD and SANAD II examined longer-term effectiveness, including treatment failure and remission. Their findings demonstrated that newer drugs were not uniformly better than established comparators. Trial interpretation also depends on seizure classification, previous treatment exposure, duration, and whether results concern monotherapy or add-on treatment. A randomized controlled trial answers a particular comparison in a defined population, not every question about a medicine.\n\nA major research objective is to move beyond seizure suppression toward prevention or modification of epilepsy itself. Antiepileptogenesis refers to preventing the development of an enduring seizure predisposition, for example after a brain insult; this has not been reliably achieved by routine prophylactic use of conventional anticonvulsants. Precision medicine instead seeks treatments matched to a specific molecular cause. Examples include metabolic therapies for defined metabolic epilepsies and pathway-directed treatments such as everolimus for tuberous-sclerosis-associated seizures. Gene-based and other targeted approaches remain active areas of investigation. Their evaluation requires outcomes that distinguish fewer seizures from improved development, altered disease progression, and durable effects after treatment ends."
    }
  ],
  "links": [
    "brain",
    "epilepsy",
    "pharmacology",
    "gaba",
    "glutamate",
    "diagnosis",
    "electroencephalography",
    "magnetic-resonance-imaging",
    "surgery",
    "health-related-quality-of-life",
    "pharmacokinetics",
    "kidney",
    "kidney-stones",
    "pancreatitis",
    "pharmacogenomics",
    "pharmacovigilance",
    "pregnancy",
    "folate",
    "stroke",
    "bipolar-disorder",
    "migraine",
    "randomized-controlled-trial",
    "precision-medicine"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}